Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto

Transcription

1 POSIVA Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto - Synhesis 2012 Posiva Oy December 2012 POSIVA OY Olkiluoto FIN EURAJOKI, FINLAND Phone (02) (nat.), ( ) (int.) Fax (02) (nat.), ( ) (int.)

2 POSIVA Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto - Synthesis 2012 Posiva Oy December 2012 POSIVA OY Olkiluoto FI EURAJOKI, FINLAND Phone (02) (nat.), ( ) (int.) Fax (02) (nat.), ( ) (int.)

3 ISBN ISSN

4 Posiva-raportti Posiva Report Posiva Oy Olkiluoto FI EURAJOKI, FINLAND Puh (31) Int. Tel (31) Raportin tunnus Report code POSIVA Julkaisuaika Date December 2012 Tekijä(t) Author(s) Posiva Oy Toimeksiantaja(t) Commissioned by Posiva Oy Nimeke Title SAFETY CASE FOR THE DISPOSAL OF SPENT NUCLEAR FUEL AT OLKILUOTO SYNTHESIS 2012 Tiivistelmä Abstract TURVA-2012 is Posiva s safety case in support of the Preliminary Safety Analysis Report (PSAR 2012) and application for a construction licence for a spent nuclear fuel repository.. Consistent with the Government Decisions-in- Principle, this foresees a repository developed in bedrock at the Olkiluoto site according to the KBS-3 method, designed to accept spent nuclear fuel from the lifetime operations of the Olkiluoto and Loviisa reactors. Synthesis 2012 presents a synthesis of Posiva Oy s Safety Case TURVA-2012 portfolio. It summarises the design basis for the repository at the Olkiluoto site, the assessment methodology and key results of performance and safety assessments. It brings together all the lines of argument for safety, evaluation of compliance with the regulatory requirements, and statement of confidence in long-term safety and Posiva s safety analyses. The TURVA-2012 safety case demonstrates that the proposed repository design provides a safe solution for the disposal of spent nuclear fuel, and that the performance and safety assessments are fully consistent with all the legal and regulatory requirements related to long-term safety as set out in Government Decree 736/2008 and in guidance from the nuclear regulator - the STUK. Moreover, Posiva considers that the level of confidence in the demonstration of safety is appropriate and sufficient to submit the construction licence application to the authorities. The assessment of long-term safety includes uncertainties, but these do not affect the basic conclusions on the long-term safety of the repository. Avainsanat - Keywords Safety case, safety assessment, synthesis, KBS-3V, Olkiluoto ISBN ISBN Sivumäärä Number of pages 277 ISSN Kieli Language ISSN English

5

6 Posiva-raportti Posiva Report Posiva Oy Olkiluoto FI EURAJOKI, FINLAND Puh (31) Int. Tel (31) Raportin tunnus Report code POSIVA Julkaisuaika Date Joulukuu 2012 Tekijä(t) Author(s) Posiva Oy Toimeksiantaja(t) Commissioned by Posiva Oy Nimeke Title TURVALLISUUSPERUSTELU KÄYTETYN YDINPOLTTOAINEEN LOPPUSIJOITUK- SELLE OLKILUODOSSA SYNTEESIRAPORTTI 2012 Tiivistelmä Abstract Posiva Oy on laatinut käytetyn ydinpolttoaineen loppusijoituslaitoksen pitkäaikaisturvallisuutta käsittelevän turvallisuusperustelun TURVA-2012 täydentämään loppusijoituslaitoksen alustavaa turvallisuusselostetta (PSAR2012) ja rakentamislupahakemusta. Eduskunnan vahvistamien periaatepäätösten mukaan Olkiluo don ja Loviisan voimalaitoksissa syntyvä käytetty ydinpolttoaine loppusijoitetaan Olkiluodon kallioperään rakennettavaan KBS-3 menetelmän mukaiseen loppusijoituslaitokseen. Synthesis raportti on synteesi TURVA-2012 turvallisuusperustelun muodostavista raporteista. Raportissa on esitetty yhteenveto Olkiluotoon rakennettavan loppusijoituslaitoksen suunnitteluperusteista, turvallisuusperustelun metodologiasta sekä toimintakykyanalyysin ja turvallisuusanalyysin keskeisimmistä tuloksista. Raportissa on niin ikään esitetty yhteenveto turvallisuutta tukevista perusteluista, arvio pitkäaikaisturvallisuuteen ja turvallisuusperustelua koskevien viranomaisvaatimusten täyttymisestä sekä arvio pitkäaikaisturvallisuuden ja Posiva Oy:n turvallisuusanalyysien luotettavuudesta. Turvallisuusperustelu TURVA-2012 osoittaa, että käytetyn ydinpolttoaineen loppusijoitus suunnitellulla tavalla on turvallista ja että toimintakyky- ja turvallisuusanalyysit vastaavat valtioneuvoston asetuksessa 736/2008 ja Säteilyturvakeskuksen YVL-ohjeissa esitettyjä pitkäaikaisturvallisuutta koskevia vaatimuksia. Lisäksi, Posiva Oy:n näkemyksen mukaan loppusijoituksen turvallisuus on osoitettu riittävän luotettavasti rakentamislupahakemusta varten. Loppusijoituksen pitkäaikaisturvallisuuden arviointiin liittyy epävarmuuksia, mutta näillä ei ole vaikutusta johtopäätöksiin käytetyn ydinpolttoaineen loppusijoituslaitoksen pitkäaikaisturvallisuudesta. Avainsanat - Keywords Turvallisuusperustelu, turvallisuusanalyysi, KBS-3V, Olkiluoto ISBN ISBN Sivumäärä Number of pages 277 ISSN Kieli Language ISSN Englanti

7

8 1 EXECUTIVE SUMMARY TABLE OF CONTENTS Spent nuclear fuel The TURVA-2012 safety case...3 Objective, audience and scope...3 Quality assurance...4 The KBS-3 method and the Olkiluoto site...4 The KBS-3 method...4 The Olkiluoto site...5 Legal and regulatory requirements...6 Design methodology...8 Requirements management...8 Safety principles, safety concept and safety functions...9 Design basis and specifications...11 Assessment methodology...12 System description...12 Features, events and processes...13 Future lines of evolution...14 Performance assessment...14 Formulation of release scenarios...15 Safety assessment...16 Results of performance assessment...18 Excavation and operation up to closure of the disposal facility Post-closure period during the next 10,000 years...19 Evolution during repeated glacial cycles...21 Uncertainties in performance assessment...24 Assessment of radionuclide release scenarios Scenarios and cases...24 Analysis of the base scenario Reference Case...25 Analysis of variant scenarios...27 Analysis of disturbance scenarios...30 Complementary analysis...31 Summary of results and uncertainties...32 Complementary considerations...34 Choice of geological disposal...34 Support for the robustness of the KBS-3 method...34 Support for the suitability of the Olkiluoto site...35 Conclusions...36 The main research and development needs during the coming years...36 Conclusions on compliance...36

9 2 Spent nuclear fuel The spent nuclear fuel that arises from the generation of electricity at the Loviisa and Olkiluoto nuclear power plants is classified as nuclear waste. According to the Nuclear Energy Act, including amendments, nuclear waste generated in Finland must be processed, stored and disposed of in Finland. Posiva Oy (Posiva) was established by Imatran Voima Oy (later Fortum Power and Heat Oy) and Teollisuuden Voima Oy in Its mission is to implement the disposal programme for spent nuclear fuel from the Loviisa and Olkiluoto nuclear power plants, and to carry out related research, technical design and development. In 2001, the Parliament endorsed a Decision-in-Principle (DiP) whereby the spent nuclear fuel produced by the operating nuclear reactors at Olkiluoto and Loviisa will be disposed of in a geological repository at Olkiluoto. Subsequently, additional DiPs were issued allowing extension of the repository to accommodate spent nuclear fuel from the operation of additional reactors that are under construction or are planned at Olkiluoto. The safety case supporting the construction licence application to be submitted by the end of 2012 foresees the disposal of spent nuclear fuel produced by these reactors during their operating lifetime, in total 9000 tu. Figure 1 provides a timeline for nuclear waste management for Olkiluoto and Loviisa reactors in which the aim is to start the disposal of spent fuel around Figure 1. Timeline for nuclear waste management relating to the Loviisa and Olkiluoto reactors until The target is to begin disposal of spent nuclear fuel around 2020.

10 3 The TURVA-2012 safety case Objective, audience and scope TURVA-2012 is Posiva s safety case in support of the Preliminary Safety Analysis Report (PSAR 2012) and application for a construction licence for a disposal facility for spent nuclear fuel at the Olkiluoto site. It is addressed to the nuclear regulator, STUK, and other national stakeholders as well as the international scientific and technical communities engaged in the discussion on nuclear waste disposal. STUK will review the safety case and related topical reports as part of its evaluation of construction licence application and the PSAR and give a statement on the construction licence application, which will form a basis for the Government judgement on issuance of the construction licence. The TURVA-2012 safety case presents the arguments for the long-term radiological safety of the planned disposal system. It includes: a description of the spent nuclear fuel to be disposed of in the geological repository; a description of the natural and engineered barrier system that the repository system provides, a definition of the safety functions and targets set for these, and a description of the present understanding of the processes that may affect the evolution and performance of the spent nuclear fuel, engineered barriers, host rock and the surface environment; a performance assessment systematically analysing the ability of the repository system to provide containment and isolation of the spent nuclear fuel for as long as it remains hazardous; a definition of the lines of evolution that may lead to failure of the canisters containing the spent nuclear fuel and to the releases of radionuclides (scenarios); analyses of the potential rates of release of radionuclides from the failed canisters, the retention, transport and distribution of radionuclides within the repository system and surface environment and the potential radiation doses to humans, plants and animals including the associated uncertainties and an evaluation of their impacts; the models and data used in the description of the evolution of the repository system and the development of the surface environment and for the analysis of activity releases and dose assessment; a range of qualitative evidence and arguments that complement and support the reliability of the results of the quantitative analyses; and a comparison of the outcome of the analyses with safety requirements. Aspects of safety related to the period of operations are dealt with in other parts of the PSAR. The TURVA-2012 safety case is presented in a portfolio of safety case reports and supporting documents, and a synthesis of these that brings together all the lines of

11 4 arguments for safety, including the main starting points, methodology, results and conclusions. Quality assurance The quality of the TURVA-2012 safety case has been assured through documented procedures in accord with Posiva s quality management principle, which is based on the ISO 9001:2008 standard. A graded approach is applied whereby the primary emphasis is on quality control of those activities that have a direct bearing on safety. The overall plan, goals and constraints for the TURVA-2012 safety case production process are presented in Posiva s Safety Case Plan The organisation of the TURVA-2012 safety case production process is referred to as SAFCA. The details of how the Safety Case Plan is being implemented are described in the SAFCA project plan. The work is managed and coordinated by a SAFCA project group and supervised by a steering group. A SAFCA quality co-ordinator has been designated for activities related to quality assurance measures applied to the production of the safety case. Improvements are made to the process as deemed useful or necessary. The quality co-ordinator is also responsible for the coordination of the expert reviews, maintenance of schedules, and review and approval of the reports. Posiva s quality manager undertakes regular auditing of the safety case production process. The KBS-3 method and the Olkiluoto site The 2001 DiP states that disposal of spent nuclear fuel shall take place in a geological repository at the Olkiluoto site, developed according to the KBS-3 method. The KBS-3 method The KBS-3 method was conceived as a solution for the disposal of spent nuclear fuel in Sweden in the early 1980s. Since then, the method has been developed and its key elements tested by SKB in Sweden and Posiva in Finland, and in joint projects. The method envisages disposal of spent nuclear fuel within a system of multiple barriers, which consists of engineered barriers and the natural barrier provided by the host rock. Posiva s reference design is based on the emplacement of canisters containing the spent nuclear fuel in vertical deposition holes (KBS-3V). Posiva is jointly with SKB developing a potential alternative design where multiple canisters are emplaced horizontally in deposition drifts (KBS-3H). The present safety case is based on the reference design. The repository is constructed on a single level with the floor of the deposition tunnels at a depth of between 400 and 450 m below the ground surface in the Olkiluoto bedrock (Figure 2).

12 5 Figure 2. Schematic illustration of the KBS-3V design. In the reference design, the spent nuclear fuel assemblies are placed into copper canisters with cast iron load-bearing inserts, and the canisters are emplaced vertically in individual deposition holes bored in the floors of the deposition tunnels. The canisters are surrounded by a swelling clay buffer material that separates them from the bedrock. The deposition tunnels, central tunnels, access tunnel and other underground openings are backfilled with materials that help to restore the natural conditions in the bedrock after operations. The Olkiluoto site The Olkiluoto site, located on the coast of south-western Finland, has been investigated as a potential site for geological disposal of spent nuclear fuel for over 25 years. This has included the construction of an underground rock characterisation facility the ONKALO. Olkiluoto Island has an area of about 10 km 2 ; the surface facilities including the encapsulation plant will occupy about 0.1 km 2 ; according to the current design and required capacity, the deposition tunnels and other tunnels will occupy about 2 km 2. The characterisation of the Olkiluoto site is focused on a volume of bedrock situated between 400 and 500 metres below ground. At this depth, favourable and predictable bedrock and groundwater conditions are found. In addition, the likelihood of inadvertent human intrusion is low.

13 6 Key features of the Olkiluoto site with respect to its suitability for geological disposal of radioactive waste include: a stable tectonic situation within the Fennoscandian Shield, away from active plate margins; good quality crystalline bedrock suitable for the excavation of self-supporting tunnels and other underground openings, such as deposition holes, technical rooms and shafts; reducing conditions at disposal depth and also otherwise favourable geochemical characteristics of the groundwater; and low groundwater flow at depth occurs currently, as it has occurred over a long period in the past, and is expected to persist for a long period into the future. The conditions in the Olkiluoto bedrock provide favourable conditions for longevity and reliable functioning of the engineered barrier system (EBS). In addition, the low groundwater flows, and physical and chemical retardation processes, limit the movement of radionuclides. Key features and processes that provide constraints on the layout of the repository and other underground openings, or that must be taken into account in the assessments of long-term performance and safety, include: presence of deformation and fractured zones, displaying more mixed geotechnical properties and in some cases increased hydraulic activity; higher rock stress at depth which may cause disturbance to the rock, making underground openings less stable; temperature and thermal conductivity of rock and residual heat output of the spent nuclear fuel; high salinity of groundwater at depth, which may affect the performance of the engineered barriers; continuing post-glacial crustal uplift and, in the longer term, climatic cooling and glaciation leading to changes in rock stress and potential changes in groundwater flow and hydrochemistry, e.g. influx of dilute glacial melt waters into the host rock. Legal and regulatory requirements The basis for the use of nuclear energy in Finland is given in the Nuclear Energy Act (YEL 990/1987) and Nuclear Energy Decree (YEA 161/1987), which came into effect in According to the Nuclear Energy Act: Nuclear waste shall be managed so that after disposal of the waste no radiation exposure is caused, which would exceed the level considered acceptable at the time the final disposal is implemented. The safe management of nuclear waste is the responsibility of the utilities that generate the waste. The responsible parties must submit reports to the Ministry of Trade and Employment every three years. These reports include a description of the measures

14 7 taken towards implementation of nuclear waste management during the last three-year period, as well as an outline of the plans for the next three years. The most recent report was submitted in September The schedule for the disposal of spent nuclear fuel was first defined by the Government in 1983 and slightly modified by the Ministry of Trade and Industry (KTM) 1 in According to the Ministry decision, the parties under the nuclear waste management obligation shall, separately, together or through Posiva Oy, present all reports and plans required to obtain a construction licence for a disposal facility for spent nuclear fuel by the end of The disposal facility is expected to become operational around Government Decree 736/2008 sets the legal requirements regarding the safety of disposal of spent nuclear fuel. The Radiation and Nuclear Safety Authority (STUK) issues guidance documents on the fulfilment of the requirements set in the Government Decree. These guides also set out STUK s expectation for the content, quality and criteria to be met by a safety case submission for disposal of nuclear waste. A total of five Guides apply to the disposal of spent nuclear fuel. The most relevant here is Guide YVL D.3, which provides guidance on the handling, storage and encapsulation of spent nuclear fuel and YVL D.5 2, which provides guidance on the planning of the disposal method, design and operation of the disposal facility, safety requirements and demonstration of compliance with safety requirements, regulatory control and on the compilation of a safety case. Guide YVL D.5 applies to disposal of all types of nuclear waste and provides guidance related to operational and long-term safety. Key requirements, stemming from GD 736/2008 and set out in Guide YVL D.5, are summarised in Table 1. The Guide provides substantial additional information on the meaning of these requirements and the evidence needed to show compliance. The Guide YVL D.5 does not specify the precise time frames over which assessments are needed. Posiva consider, however, that radiation doses can be assessed, assuming human habits, nutritional needs and metabolism remain unchanged, with sufficient reliability over a period of up to 10,000 years, and that the fulfilment of the safety functions of the repository system and the release of radionuclide to the surface environment can be reasonably assessed up to one million years after repository closure. Table 1. Synthesis of key requirements for long-term safety from STUK s Guide YVL D.5. Please refer to the Guide for actual wording and context. Related to long-term radiological impacts For expected evolution scenarios, and in the period during which the radiation exposure can be assessed with sufficient reliability (at least over several millennia): the annual dose to the most exposed people shall remain below the value of 0.1 msv; the average annual doses to other people shall remain insignificantly low. In the longer term, the radiation impacts arising from disposal can at a maximum be equivalent to those arising from natural radioactive substances in Earth s crust, and on a large scale should remain 1 2 Now Ministry of Trade and Emplyoment The Guides YVL D.3 and YVL D.5 is are available in draft form. STUK has agreed that the licence application can be based on draft version 4 of both Guides (version has been used).

15 8 insignificantly low. The nuclide-specific constraints on releases to the environment (average release of radioactive substances per annum) are specified in YVL D.5. For the activity releases that arise from the expected evolution scenarios, the sum of the ratios between the nuclide-specific activity release rates and the respective constraints given in YVL D.5 shall be less than one (this is evaluated through the release rates for radionuclides from the geosphere to the biosphere), The importance of unlikely events impairing long-term safety shall be assessed, and whenever practicable, the radiation impacts caused shall be assessed quantitatively. The resulting annual radiation dose or activity release shall be calculated and multiplied by its estimated probability of occurrence. The obtained expectation value shall be below the dose constraint (see above) or release constraints given in Table 2-4. The assessed radiation exposures to fauna and flora shall remain clearly below the levels that could cause decline in biodiversity or other significant detriment to any living population. Related to providing long-term safety Disposal shall be implemented in stages, with particular attention paid to aspects affecting long-term safety. The long-term safety of disposal shall be based on safety functions achieved through mutually complementary barriers so that a deficiency of an individual safety function or a predictable geological change will not jeopardise the long-term safety. Targets shall be specified for the performance of each safety function based on high quality scientific knowledge and expert judgement. For spent fuel, the safety functions provided by the engineered barriers shall limit effectively the release of radioactive substances into bedrock for at least 10,000 years. The characteristics of the host rock shall be favourable for the long-term performance of engineered barriers and with respect to the groundwater flow regime at the disposal site. Design methodology Requirements management Posiva has developed a robust design for geological disposal of spent nuclear fuel at Olkiluoto through a formal requirements management system (VAHA). This provides a rigorous, traceable method of translating safety principles and the safety concept to a set of safety functions, performance requirements, design requirements and design specifications for the various barriers, i.e. a specification for enactment of the disposal concept at the Olkiluoto site. The VAHA sets out: At Level 1, stakeholder requirements that come from laws, decisions-in-principle, regulatory requirements, and other stakeholder requirements; At Level 2, the long-term safety principles, which lead to the definition of the safety concept and safety functions; At Level 3, the performance requirements, consisting of performance targets for the engineered barriers and target properties for the host rock, such that the safety functions are fulfilled; At Level 4, the design requirements for the engineered barriers and the underground openings including rock suitability classification criteria (RSC criteria), such that the performance requirements will be met; At Level 5, the design specifications, which are the detailed specifications to be used in the design, construction and manufacturing.

16 9 Safety principles, safety concept and safety functions The long-term safety principles set out for the KBS-3 method are based on the use of a multi-barrier disposal system consisting of engineered barriers and host rock. The engineered barrier system consists of the canister, buffer, backfill of the deposition tunnel and closure. The role of the engineered barriers is to provide the primary containment against the release of radionuclides. The host rock should provide favourable conditions for the long-term performance of the engineered barriers, but also limit or retard the transport of radionuclides. The multi-barrier system as a whole should be able to protect the living environment even if one of the barriers turns out to be deficient. The safety concept (Figure 3) is a conceptual description of how these principles are applied to achieve safe disposal of spent nuclear fuel in the present-day and future conditions of the Olkiluoto site. Containment of the radionuclide inventory associated with the spent nuclear fuel is provided first and foremost by encapsulating the fuel in sealed (gas-tight and watertight) copper-iron canisters. The other EBS components (buffer, backfill and closure) provide favourable near-field conditions for the canisters to remain intact and, in the event of canister failure, slow down and limit releases of radionuclides from the canister. The containment of radionuclides is ensured by the proven technical quality of the EBS. Other elements of the safety concept include sufficient depth for the repository, favourable and predictable bedrock and groundwater conditions and well-characterised material properties of both the bedrock and the EBS. A robust system design ensures that single deficiencies in the design or implementation of the design, or uncertainties in future conditions, do not lead to significant weakening of the overall safety of the repository system. Safety functions are assigned to the components of the engineered barrier system (EBS) and the host rock as shown in Table 2. Most of the activity in the spent nuclear fuel is contained in a ceramic matrix (UO 2 ) that is resistant to dissolution in the expected repository conditions. The slow release of radionuclides from the spent fuel matrix in the event of canister failure is part of Posiva s safety concept. However, no safety functions or performance requirements (see below) are assigned to spent nuclear fuel; rather, the properties of the spent fuel are used as a starting point in the design of the disposal system. Posiva is responsible for the disposal of all spent nuclear fuel from Olkiluoto and Loviisa power plants, and if the current design of the disposal system would not provide of a sufficient level of safety for disposal of a possible new specific fuel type, the design will be modified to meet the safety requirements.

17 10 SAFE DISPOSAL LONG-TERM ISOLATION AND CONTAINMENT FAVOURABLE NEAR-FIELD CONDITIONS FOR THE CANISTER Retention and retardation of radionuclides Slow transport in the geosphere Slow release from the spent fuel matrix Slow diffusive transport in the buffer PROVEN TECHNICAL QUALITY OF THE EBS FAVOURABLE, PREDICTABLE BEDROCK AND GROUNDWATER CONDITIONS SUFFICIENT DEPTH WELL-CHARACTERISED MATERIAL PROPERTIES ROBUST SYSTEM DESIGN Figure 3. Outline of the safety concept. Orange blocks indicate the primary safety features and properties of the disposal system. Green blocks indicate secondary safety features that become important in the event of a radionuclide release from a canister. Table 2. Safety functions assigned to the barriers (EBS components and host rock) in Posiva s KBS-3V repository. Barrier Safety functions Canister Ensure a prolonged period of containment of the spent nuclear fuel. This safety function rests first and foremost on the mechanical strength of the canister s cast iron insert and the corrosion resistance of the copper surrounding it. Buffer Contribute to mechanical, geochemical and hydrogeological conditions that are predictable and favourable to the canister. Protect canisters from external processes that could compromise the safety function of complete containment of the spent nuclear fuel and associated radionuclides Limit and retard radionuclide releases in the event of canister failure. Deposition tunnel backfill Contribute to favourable and predictable mechanical, geochemical and hydrogeological conditions for the buffer and canisters. Limit and retard radionuclide releases in the possible event of canister failure. Contribute to the mechanical stability of the rock adjacent to the deposition tunnels. Host rock Isolate the spent nuclear fuel repository from the surface environment and normal habitats for humans, plants and animals and limit the possibility of human intrusion, and isolate the repository from changing conditions at the ground surface. Provide favourable and predictable mechanical, geochemical and hydrogeological conditions for the engineered barriers. Limit the transport and retard the migration of harmful substances that could be released from the repository. Closure Prevent the underground openings from compromising the long-term isolation of the repository from the surface environment and normal habitats for humans, plants and animals. Contribute to favourable and predictable geochemical and hydrogeological conditions for the other engineered barriers by preventing the formation of significant water conductive flow paths through the openings. Limit and retard inflow to and release of harmful substances from the repository.

18 11 Design basis and specifications The definition of the performance targets for the safety functions of the engineered barriers and the target properties for the safety functions of the host rock requires the identification of the different loads and interactions that may act on the repository system at the time of canister emplacement and in the long term. To achieve this, the potential future conditions have to be described as alternative lines of evolution, and their likelihoods are assessed based on present-day understanding and the findings of earlier assessments. All the lines of evolution and expected loads that are judged reasonably likely to occur (based on this understanding and previous findings) are taken into account and, hence, included in the design basis. Thus, by definition, when the performance targets and target properties are met and the future follows the reasonably likely lines of evolution (design basis scenarios), the safety functions are fulfilled. From the performance targets and target properties (VAHA level 3) the design requirements are derived (VAHA level 4). Then, design specifications are worked out such that the fulfilment of these requirements can be verified at implementation (VAHA level 5). Performance assessment shows that the system, as designed and built according to the design requirements and specifications, will meet the performance targets and target properties and thus that the safety functions will be fulfilled for an envelope of future conditions that includes all reasonably likely lines of evolution. In defining the performance targets for the engineered barriers, implementation aspects have to be considered. The performance targets have to be set considering, on the one hand, the long-term safety aspects and, on the other hand, that the design and implementation must be robust, as that is the foundation of the safety concept. For the rock barrier, the target properties set the starting point for the definition of the Rock Suitability Classification system (RSC) developed by Posiva. The classification system includes both the updated rock suitability criteria as well as the procedure for the suitability classification during the construction of the repository. The RSC is used to identify suitable rock volumes for repository panels and to assess the suitability of deposition tunnels for locating deposition holes and to accept deposition holes for disposal. The performance targets and target properties, together with the derived design requirements and the underlying design basis scenarios, form the design basis of the repository. The background and premises for the design basis are presented in Design Basis. A repository system designed and built according to the design basis is expected to comply with the regulatory safety requirements. However, the safety case also includes a discussion of situations in which the system does not meet all the requirements, where there are uncertainties in whether they are met, or if the system follows an unlikely line of evolution.

19 12 Assessment methodology Figure 4 outlines the approach to the development of the safety case, whereby the design basis is developed, the performance of the repository system assessed, and scenarios leading to radionuclide release are formulated and assessed. The design basis and definition of performance targets and target properties are developed iteratively between performance assessment, formulation and assessment of radionuclide release scenarios and presentation of the safety case. Available scientific understanding, including the results from earlier assessments, is used in the definition of the performance targets, target properties for the host rock, design requirements and criteria for rock classification. These will be updated as scientific understanding is further developed, taking into account the results of the performance assessment and assessment of radionuclide release scenarios of the current safety case (the two-way arrows in Figure 4). System description An accurate and reliable description of the disposal system is the foundation both for the development of a robust design (previous section), and for an understanding of the possible lines of evolution of the disposal system, assessments of performance and safety, and complementary considerations that comprise the safety case. The results of the safety case are used to specify the further development of the disposal facility, if needed. Characterisation studies of the Olkiluoto site have been made for over 25 years. This has lead to a detailed description and understanding of the site in respect of all characteristics relevant to the construction of a repository for spent nuclear fuel and to its long-term evolution. Studies of the surface environment of the site form the basis for a description of the biosphere sufficient to characterise the environment to be protected and its potential future use and occupation by humans, plants and animals. Descriptions of the site and surface environment are provided in Site Description and Biosphere Description. The KBS-3 method and the KBS-3V design have been developed over more than 30 years. The specific realisation of the design as planned for implementation of a repository at the Olkiluoto site is the result of thorough analyses of the functional requirements of the engineered barriers and host rock and of the overall safety of the repository system. Detailed descriptions of the components of the repository system and evidence concerning their practical realisation and feasibility have been compiled in production line reports. The main characteristics, initial state including uncertainties of the repository system components (spent nuclear fuel, EBS and host rock) and of the surface environment to be used as input to the safety assessment have been compiled in Description of the Disposal System.

20 13 Figure 4. Approach to the development of the safety case (FEP= Features, Events and Processes, PSAR= Preliminary Safety Analysis Report, FSAR = Final Safety Analysis Report). Features, events and processes Identifying and describing the features, events and processes (FEPs) that are relevant to the evolution of the disposal system, or to its potential performance and safety, is an essential step towards ensuring comprehensiveness of the assessments and safety case. For the TURVA-2012 safety case, the identification and screening of FEPs has been carried out by a team of scientific subject and assessment experts, based on a review of the FEPs considered in Posiva s previous assessments, the NEA FEP-database and the FEPs considered in safety cases in other nuclear waste programmes, as well as an examination of the specific characteristics of the disposal system and the Olkiluoto site. A FEP database has been developed providing a structured classification of relevant FEPs and couplings between these. The FEPs are presented in Features, Events and Processes, including a description of each FEP and the fundamental uncertainties based on current scientific understanding. The relevance of each FEP for the long-term safety of the disposal facility to be constructed at Olkiluoto has been evaluated based on the situations in which the FEP could occur at the Olkiluoto site and its couplings to other FEPs. Spent nuclear fuel must be kept isolated for as long as it could cause significant harm to the normal habitats for humans, animals and plants. In TURVA-2012 an assessment time frame of up to one million years into the future is considered. This is consistent with other assessments of spent nuclear fuel disposal internationally. At one million

21 14 years, the activity of the spent nuclear fuel is similar to that of the original uranium ore from which the fuel was fabricated. Future lines of evolution The understanding of FEPs is used to develop descriptions of future lines of evolution of the repository system (the engineered barriers and host rock) and of the surface environment. This provides the framework for estimating the thermal, hydraulic, mechanical and chemical (THMC) loads that will be placed on the system. During the construction and operation of the repository up to its closure, the main changes are related to excavation effects and draining of water from the underground openings, plus introduction of radiogenic heat from the spent nuclear fuel. Some limited mechanical damage immediately around the openings, as well as an increase of groundwater flow into the repository volume and changes in hydrochemistry are expected. After closure, groundwater flow will return towards preconstruction conditions, although modified by radiogenic heat from the spent fuel for a time. Salinity will be reduced in the longer term due to infiltration of meteoric waters. In the longer term, the main driver for change is climate evolution, where the expected case is a continuation of glacial-interglacial cycling as experienced over the last one million years of the Quaternary. However, best scientific understanding indicates the past and continuing anthropogenic emissions of CO 2 and other greenhouse gases will lead to increased global temperatures over a period of many thousands of years, delaying the onset of cooler climate conditions. Thus, over the next 50,000 years, conditions are expected to remain essentially as today, i.e. a temperate climate with a boreal ecosystem. A first cold period is not expected until about 50,000 years after present (50 ka AP) with temperature and precipitation changes leading to permafrost development and, later on, to ice-sheet development. For the assessment, from 50 to 170 ka AP, a repetition of the sequence of events during the last glacial cycle is assumed. After 170 ka AP, seven repetitions of the cycle from 50 ka to 170 ka AP are assumed. Thus, a total of eight glacial cycles are accounted for in the assessment time frame (up to one million years after present). Variations in the duration and intensity of individual glacial cycles are not expected to have a significant impact on repository safety. Performance assessment Performance assessment shows that the system, designed and built according to the design requirements and specifications, is compliant with the performance targets and target properties initially and in the long term and that the safety functions will be fulfilled (Table 2), which, being so, will lead to isolation of the spent nuclear fuel, and complete containment over hundreds of thousands of years and even for the one million year time frame. The performance of the repository system is analysed and the fulfilment of performance requirements is evaluated taking into account the expected thermal, hydraulic, mechanical and chemical evolution of the repository system, and uncertainties in the expected lines of evolution. Less expected lines of evolution, including the possibility of disruptive events, are also identified. Account is taken of the natural evolution of the environment, chiefly driven by climatic evolution, which imposes external loads on the

22 15 repository system, and also internal loads, chiefly from the effects of excavation and emplacement of the spent nuclear fuel and the engineered barriers. The performance is considered below in three time windows: (1) during the excavation and operational period up to closure; (2) up until 10,000 years after closure; (3) beyond 10,000 years over repeated glacial cycles. The fulfilment of performance targets and target properties in each time window is assessed considering time-dependent and space-dependent loads on the engineered barriers and host rock. Quantitative assessments are made whenever possible, e.g. to calculate safety margins and demonstrate the robustness of the design. Uncertainties are highlighted, conditions that could lead to deviations from performance targets and target properties are identified, and the likelihood and effects of the deviations estimated whenever possible. In particular, conditions and events (incidental deviations) that could lead to the release of radionuclides are identified; these are taken forward to the formulation of radionuclide release scenarios and to the radiological impact assessment of radionuclide releases. Formulation of release scenarios Consistent with the regulatory guidance (Table 1 and Guide YVL D.5), Posiva distinguishes between the expected evolution of the disposal system and unlikely evolutions and events. The repository system is designed in a way that for the expected lines of evolution of the system, each component of the EBS meets the performance targets assigned to it, and the host rock conforms to its target properties. In this case the copper canisters (with iron inserts) remain intact for the whole assessment time frame and there is no release of radionuclides. This is confirmed in performance assessment. Performance assessment shows, however, that there are some plausible conditions and (incidental deviations) that could lead to the reduction of one or more safety functions, and thus may give rise to radionuclide releases. In addition, there are some unlikely events and processes that could disrupt the repository, e.g. related to human intrusion and rock shear. These incidental deviations and unlikely events are systematically examined to define a set of scenarios that encompass the combinations of initial conditions, evolution and disruptive events. In the current and past assessments by Posiva, the scenario of a canister with an initial penetrating defect has been considered to test the radiological performance of the other engineered barriers and host rock. This defect is most likely in the weld. Although the likelihood that a canister with an initial undetected penetrating defect will be emplaced is low, this is a useful base scenario for safety assessment (radionuclide release calculations) against which the efficiency of the other engineered barriers and the host rock to limit the radionuclide releases can be tested and that also complies with GD 736/2008. The classification of scenarios in TURVA-2012 is illustrated in Figure 5. The base scenario addresses the most likely lines of evolution (in which the performance targets and safety functions are met), but takes into account the possibility of the emplacement

23 16 Figure 5. Classification of scenarios in TURVA-2012, which is consistent with STUK s Guide YVL D.5. of one or a few canisters with initial undetected penetrating defects. The variant scenarios address situations that are considered reasonably likely and in which there may be reduced performance of one or more safety functions of the barriers. Disturbance scenarios address the lines of evolution that are considered unlikely but cannot be completely eliminated. Safety assessment The aim of safety assessment is to analyse the radionuclide release, transport and radiological impacts of the identified scenarios, and scenario combinations, and to compare the calculated impacts with regulatory criteria in order to judge their acceptability. In the evaluation of the releases due to unlikely scenarios, the likelihood of the assumed conditions or events is considered. The main safety indicators calculated in TURVA-2012 are the following. 1. The radioactive releases from the bedrock to the biosphere (surface environment), which are calculated for all release scenarios and assessed against the nuclidespecific constraints for the radioactive releases to the environment (average annual release rates of radioactive substances) defined in YVL D Annual doses 3 to humans. Consistent with regulatory guidance (Table 1) these are calculated for scenarios that give rise to releases to the surface environment in the first 10,000 years. 3. Absorbed dose rates to plants and animals calculated for releases to the surface environment in the first 10,000 years. The repository system is analysed using models that represent: release from the spent nuclear fuel (taking account of the locations of radionuclides in the fuel, its cladding and other parts of the fuel element); 3 In this report, annual dose refers to the sum of the effective dose arising from external radiation within the period of one year, and the committed effective dose from the intake of radioactive substances within the same year (GD 736/2008). Furthermore dose refers to effective dose, unless otherwise explicitly stated.

24 17 release, retention and transport in the near field (release from the canister, migration through the buffer, migration by alternative routes to water-conducting fractures in the host rock); and retention and transport in the geosphere (through water-conducting fractures taking account of variability in flow paths). This yields radioactive releases from the geosphere to the biosphere, which are used as input to biosphere models. Since the repository system models are run independently of the biosphere models, the output from a single repository system calculation can be input to alternative biosphere models so as to represent alternative surface environment conditions at the time of release. Modelling for biosphere assessment includes, first, a screening process to identify those radionuclides that could make significant contributions to the total radiological impact. These radionuclides are carried forward to detailed biosphere modelling, based on a model of the future landscape and ecosystem development in the Olkiluoto area over the next 10,000 years. This provides the framework for modelling of radionuclide movements within compartments of the future surface environment and calculation of the radiation doses to humans, plants and animals, inhabiting or making use of the various areas and resources that may become contaminated. Calculation cases analyse the radiological impacts and illustrate the impact of specific uncertainties or combinations of uncertainties. These are uncertainties related to the scenario definitions, alternative model representations and data used in the models. Four types of calculation cases are distinguished: A Reference Case is one model realisation of the base scenario. Models and data for the Reference Case are, in most instances, selected to be either realistic or moderately cautious, i.e. radiological impacts are not to be underestimated nor excessively overestimated. Sensitivity cases represent alternate models and/or data to those of the Reference Case, but remain within the scope of the base scenario and/or variant scenarios. Analyses of the sensitivity cases illustrate the effect of model and data uncertainties. What-if cases are mainly model representations of disturbance scenarios. Models and data for these what-if cases are selected to represent unlikely events and processes. Complementary cases are designed to develop a better understanding of the modelled system or subsystems and to test robustness. All of the above cases are analysed deterministically, i.e. calculations are carried out for a specific set of input parameter values. In addition, the disposal system behaviour is explored by Monte Carlo simulations and probabilistic sensitivity analysis (PSA). This involves performing a set of calculations in which the values of input parameters are selected randomly from specified probability density functions (PDFs) that represent the uncertainty in each parameter. The outcomes of multiple simulations are assembled into a probability distribution and the sensitivity of outcome to inputs is investigated by statistical techniques.

25 18 Results of performance assessment Excavation and operation up to closure of the disposal facility In the period of excavation and operation, groundwater flow modelling indicates an increase of approximately two orders of magnitude in flow rates in the rock volume surrounding the repository from a pre-construction baseline. Following installation of tunnel and shaft backfill and seals (i.e. after closure), modelled flow rates return to near pre-excavation rates; however, a few deposition holes with flow rates and transport resistances outside the range defined by the target values may remain. During the open period, the average salinity around the repository remains similar to the pre-construction phase, but increased groundwater flow into the repository volume may lead to mixing of water and either more dilute or more saline conditions locally at the repository depth. The disturbed conditions are related to the main hydrogeological zones and the ONKALO facility, not necessarily to the repository panels themselves. Moreover, the disturbed conditions are likely to last a limited time; in the order of tens of years, and thus the impact on the performance of the buffer and backfill is limited. The changes in groundwater salinity remain consistent with target properties, meaning that the buffer and backfill functions are preserved in most deposition holes. Calculations of temperature evolution show a maximum temperature at the canister surface of 95 C assuming an unsaturated buffer and 75 C for a saturated buffer. The maximum rock temperature at the deposition hole wall is about 65 C at 40 years after emplacement. Thus, temperatures will remain within the performance targets. Excavation will cause a damaged zone (EDZ) to form, especially below the tunnel floors, although the damage is probably not continuous. In addition, excavation and the heat produced by the spent nuclear fuel may cause spalling or other types of stressinduced damage around the excavated openings. The uncertainties concerning the properties of the EDZ and the rock damage around the deposition holes are taken account of in groundwater flow modelling. Before full saturation, some buffer and backfill material may be lost through piping and erosion. Based on calculated inflows to deposition holes, some limited buffer loss is expected in roughly one third of the positions, but in all cases the average buffer density remains consistent with the performance target, so that the necessary low hydraulic conductivity and sufficient swelling pressure will be achieved as the buffer saturates. It is estimated that 13,000 kg of the backfill material could at most be lost locally by piping and erosion, and redistributed within the deposition tunnel. This is rather small compared with the total mass of backfill material in the tunnel (more than 8000 tonnes in a 300 m long deposition tunnel). The effect on the backfill performance depends on how the mass loss is distributed in the backfill. Backfill loss will be largest in the vicinity of fractures with a high enough inflow to transport the mass further down the tunnel. However, this type of erosion would not be detrimental for the EBS as no deposition holes would be located near such a fracture. Thus, the buffer and backfill will remain consistent with their performance targets even considering the process of piping and erosion.

26 19 For both unsaturated and saturated conditions, the consumption of oxygen in the backfill and buffer will be relatively rapid, due to its reaction with pyrite and other accessory minerals. Thus, anoxic, reducing conditions will be quickly established around the emplaced canisters and throughout the buffer and backfill. Cementitious leachates from grouting of fractures, from grout used to stabilise rock bolts and from the plug in the deposition tunnel may locally affect the backfill. However, no cement is in direct contact with the buffer and the flux of cementitious leachates reaching the buffer is estimated to be of little significance. The maximum corrosion depth of the copper canisters from the atmospheric and initially trapped oxygen is expected to be less than 0.5 mm. A probabilistic analysis has been carried out to assess the potential number of defective canisters that could be placed in the repository. The conclusion was that the data currently available are not enough for a statistical evaluation of the probability that the penetrating defects are detected before emplacement. Consequently, the probability of detection can only be based on expert judgement, taking account of the results from both the non-destructive testing and destructive testing of the weld. With more data becoming available in the future, it is likely that it will be possible to demonstrate that the probability of emplacing more than one canister with an initial undetected penetrating defect is less than one per cent. In summary, the properties of the EBS and host rock will conform to the performance targets and target properties at the end of the operational period, with some possibility of incidental deviations: an undetected penetrating defect in one or a few canisters, higher flow rate or lower transport resistance than the target values for a few deposition holes and groundwater composition outside the target range for a short time during operation and soon after closure for a few deposition holes. Post-closure period during the next 10,000 years Over the next 10,000 years, the climate is expected to remain essentially as today, i.e. a temperate climate with a boreal ecosystem. Groundwater flow and chemistry will recover from the disturbances caused by the excavation, and will slowly evolve as a response to naturally occurring gradients. Key processes during this period will be water uptake, swelling and homogenisation of the clays in the buffer, backfill and seals, and the decline of the residual heat from the spent nuclear fuel. Crustal uplift will continue but at gradually lower rates, and higher hydraulic gradients will develop close to the shoreline. At 1000 to 2000 years after present, the shoreline will have retreated far enough that further changes will not affect the flow rates in the repository volume. The heat from the spent nuclear fuel increases the flow rates at the repository depth by a factor of 2 to 3 compared with the natural state during the first hundreds of years, and enhances temporarily and locally upward flows. The heat tends to result in an upward driving force for the water, but when combined with the stronger natural downward forces, the flow remains mainly directed downwards. Heat production declines to very low levels after the first few thousands of years, and the flow returns to its natural state.

27 20 Groundwater flow modelling based on a discrete fracture network approach provides information about the migration paths and flows around the deposition holes. This information is used to assess whether the target properties are met as well as to provide input to the radionuclide release and transport analysis. The modelling studies quantify the effect of the tunnel EDZ and the rock damage around the deposition holes (including thermally induced damage) on local flow rates and other flow-related transport parameters. The presence of the damaged zone increases the connectivity of fractures and flow around the deposition hole, but the effects on the natural fractures are limited, and flow rates in natural fractures and the transport resistances in the vicinity of the deposition holes are consistent with target properties for most deposition holes. Groundwater flow modelling and subsequent reactive transport modelling show that the salinity field at the repository depth recovers from repository excavation, but at a much slower rate than the flow field. The natural salinity state is reached within hundreds of years. As the disturbances caused by repository construction cease, the groundwater composition stabilises and the variation seen during the operational period diminished. The few local values that were outside the target value range return within the range in a relatively short time. At repository depth, the ph remains close to 7.5 and reducing conditions prevail. In the longer term, salinity, chloride concentration and total charge equivalent of cations all decrease very slowly, due to the infiltration of meteoric water, but the concentrations remain consistent with the target values over the time window in question. Groundwater flowing into the repository leads to saturation and swelling of the buffer and backfill. Initial differences in the density and swelling pressure will be evened out (homogenisation), although some heterogeneity will remain. The time to reach full saturation in the buffer is calculated as a few tens to several thousands of years, depending on the local hydraulic conditions. Calculations show that expansion of the buffer into the backfill and the changes in the density of the buffer will not be sufficient to threaten the performance targets for the buffer and backfill (i.e. a sufficiently high density will be maintained). While heat is generated by the spent nuclear fuel, the thermo-hydro-mechanicalchemical evolution will lead to geochemical changes in the buffer, but these will be limited. After saturation and development of the full swelling capacity, the changes will be even lower, constrained by diffusive processes. In particular, no or only minor clay mineralogical changes will occur. The production of sulphide via microbial processes in the buffer will be minor. Further, the already minor impact of cementitious leachates on the buffer is estimated to diminish. The evolution of porewater chemistry in the backfill will be similar to that in the buffer, but will be less affected by the heat from the spent fuel. Thermally-induced clay alteration and cementation will be negligible. Disturbances due to leachates from cement materials will diminish in general and also locally, due to the low concentrations of alkalis in the leachates. Production of sulphide via microbial reduction of sulphate cannot be ruled out in localised zones of low backfill density; this is accounted for in the analysis of canister corrosion.

28 21 There are no major uncertainties in the evolution of the closure components during the first 10,000 years after closure. Even if the hydraulic plugs degrade, no preferential paths will form. At depth, transport through closure components will still be dominated by diffusion. Sulphide is the main agent for canister corrosion. Microbially produced sulphide in the buffer is negligible in this period; sulphide supply from the backfill is limited by the precipitation of iron sulphide and losses to the rock mass. Moreover, the sulphide has to diffuse through a thick layer of bentonite to reach the canister. Corrosion calculations coupled with groundwater flow modelling, and taking account of the possibility of early buffer erosion, show that the total corrosion depth will be negligible during the first 10,000 years. The initially intact canisters will remain intact for all conceivable loads that could occur during the first 10,000 years and thus the spent nuclear fuel remains contained within the canister. In summary, the properties of the EBS and host rock will conform to the performance targets and target properties over the period up to 10,000 years, with some possibility of incidental deviations: an undetected penetrating defect in one or a few canisters; higher flow rate or lower transport resistance than the target values for a few deposition holes and groundwater composition outside the target range for a short time during repository operation and soon after closure for a few deposition holes; and local lower density areas in the backfill where there is the possibility that sulphate reduction may occur. Evolution during repeated glacial cycles In the longer term, major climatic changes are expected, as described under Future lines of evolution above. Effects include permafrost, glaciation and associated sea-level changes. These changes affect the isostatic load, rock stresses, and groundwater flow and composition, as well as the mechanical and thermal evolution of the EBS and host rock. During the continued temperate climate up to 50,000 years AP, there is a slight increase in the groundwater flow rates in the upper part of the bedrock, due to surface environment changes. The flow rates at repository depth are not significantly affected. The continuing infiltration of meteoric water results in slowly decreasing salinity so that, towards the end of this period, a few canister positions may experience dilute conditions. Groundwater flow and salinity have been modelled for two representative periods of permafrost development, during which permafrost reaches depths of about 80 m and 300 m. The effects of an ice sheet have also been modelled considering an immobile ice sheet over the whole of Olkiluoto Island for 1000 years, and a retreating ice sheet. Under permafrost conditions, the hydraulic conductivity in the rock is reduced by several orders of magnitude and the infiltration is very low. As a result, the groundwater salinities remain at the level prevailing before the onset of the permafrost. During ice-sheet retreat, the flow rates through the repository volume depend on the location of the ice margin with respect to the repository. While the repository is still below the ice sheet but the ice margin is close, the flow rates are significantly increased

29 22 (by a factor of 4 to 7) and directed downwards. As the ice passes the site, the main flow direction is upwards and flow rates reduce as the distance to the ice margin increases. Some canister locations might then experience higher flow rates and lower transport resistances than the target properties. This has been taken into account in the assessment of the canister corrosion rates and in the formulation of release scenarios. Nevertheless, for most of the deposition holes, the host rock target properties related to groundwater flow are fulfilled during ice-sheet retreat. Although there is no evidence that fresh meltwater ever reached repository depth at Olkiluoto during the last glacial cycle or in the previous ones, dilute conditions around some of the deposition holes during a future ice-sheet retreat phase might be possible depending on the duration of meltwater infiltration, which could give rise to chemical erosion of the buffer and backfill (see below). Other geochemical properties (ph, redox conditions, chloride concentration, total charge equivalent of cations, sulphur and iron species) are all expected to remain consistent with the target properties throughout the period, including during ice-sheet retreat and melting. Oxygen will be consumed within short distances along the flow path and thus not reach the repository level. Although groundwater data clearly indicate sulphide values below 1 mg/l, a pessimistic upper bound of 3 mg/l is adopted in corrosion calculations described below; this accounts for possible solubility control by the more soluble amorphous iron sulphide in combination with kinetically constrained limited availability of iron and the uncertainties related to the microbial activity and availability of nutrients and energy sources. The possibility of a large earthquake leading to secondary shear movements on fractures intersecting deposition holes and to canister failure, especially at a time of glacial retreat, cannot totally be excluded. The risk of canister failures due to secondary shear movements in the event of a large earthquake can be reduced by locating the deposition holes away from large deformation zones and by avoiding large fracture intersections in deposition holes, but it is estimated that few tens of canisters may still be in positions such that they could potentially fail in such an event over a one million year time frame. On the other hand, the average annual probability of an earthquake large enough potentially to lead to canister failure due to secondary movements on fractures is, estimated to be low, in the order of This is based on the frequency of occurrence of earthquakes in the Olkiluoto area and the fact that there are around five fault zones within and around the area of the repository that could host such an earthquake. Thus, during the first glacial cycle, there is little likelihood of canister failure due to rock shear, although the possibility of such failures cannot be discounted over a one million year time frame. Freezing of the buffer or the deposition tunnel backfill is not an issue because, based on evidence from the past, permafrost will not reach the repository depth. In any case, the buffer and backfill would withstand the freeze/thaw cycles without damage to their safety functions.

30 23 The evolution of porewater salinities in the buffer and backfill will follow those in the surrounding groundwaters, which will remain within the required target ranges, except perhaps for short times during ice-sheet retreat and melting period. Under these conditions, dilute groundwater conditions might cause some chemical erosion of buffer and backfill. With the reference assumptions on groundwater flow (a selected realisation of the DFN flow model) and evolving groundwater composition, one canister position is calculated to undergo buffer erosion during the first glacial cycle to an extent that advective conditions arise. This calculation should be seen as illustrative, being based on only a single realisation of the DFN groundwater flow model. An analysis of statistical distributions of flow-related parameter values between canister positions shows that measures such as the mean and 90th percentile vary little between DFN realisations. However, the number of canister positions experiencing advective conditions is determined by the tails of these distributions, and is therefore subject to more uncertainty. Taking a more cautious view on this and other uncertainties, buffer erosion might result in advective conditions in a few canister positions. The backfill in parts of the central tunnels may lose clay components due to chemical erosion, but this will not jeopardise the overall safety functions of closure. Degradation of closure plugs is uncertain, but the swelling clays used in the lower parts of the tunnels and shafts will ensure sufficient isolation capacity of the sealing structures. As at earlier times, sulphide is the main agent for corrosion of the copper canisters. Calculations of the corrosion depth in one million years have been made assuming that the buffer performs as designed, a pessimistic sulphide concentration of 3 mg/l and a range of flow conditions. The results show that the overall corrosion depth will not exceed few tenths of a millimetre even over one million years. Thus, if the buffer performs as designed, no canister failures due to corrosion are expected even with high sulphide concentrations. Furthermore, even if the buffer is affected by chemical erosion, few if any canister failures due to corrosion are expected during the first glacial cycle, as long as conditions otherwise correspond to the expected evolution (i.e. performance targets and target properties are met). The calculated rate of corrosion and the calculated number of canister failures in these circumstances depends on the assumptions made about groundwater flow and composition, corrosion area, fracture apertures, the rate of buffer erosion and the possibility of locally thinner parts of the copper overpack. Cautiously assuming a sulphide concentration of 3 mg/l in the groundwater, but with realistic assumptions concerning these other factors, chemical erosion of the buffer and subsequent corrosion by sulphide is calculated to lead to no canister failures within the first glacial cycle, and 4 5 failures in the million year time frame. Based on more cautious assumptions, around 3 canister failures are calculated to occur within the first glacial cycle, and a few tens of failures in the million year time frame. In summary, after the first glacial cycle, i.e. more than 100,000 years after repository closure, the properties of the EBS and host rock will still conform to the performance targets and target properties, with some incidental deviations: an undetected penetrating defect in one or a few canisters, higher flow rate or lower transport resistance than the target values for a few deposition holes, erosion of buffer in some deposition holes due to long-term infiltration of meteoric water or dilute glacial meltwater and canister failure by corrosion due to unfavourable groundwater conditions and buffer erosion and canister failure due to shear displacements in fractures during ice-sheet retreat.

31 24 Successive glacial cycles will impose similar loads as considered during the first glacial cycle. Thus, over the one million year assessment time frame: the potential for buffer erosion increases for deposition holes that experience dilute groundwater conditions during ice-sheet retreat; the number of deposition holes that suffer a shear displacement sufficient to cause canister failure could increase; the extent of canister corrosion in deposition holes that suffer buffer erosion could increase. Over the one million year time frame, the properties of the EBS and host rock will still conform to the performance targets and target properties except for the incidental deviations listed above. Uncertainties in performance assessment If the engineered barriers and the rock for the whole repository system fulfil the set performance requirements, no releases are expected during more than 100,000 years after closure. However, deviations from performance targets and target properties (Table 3) may lead to radionuclide releases. The importance of these releases has been assessed considering different radionuclide release scenarios. Table 3. Summary of deviations from performance targets and target properties as may occur and are relevant in each time window. Deviations Up to closure of the disposal facility Up to 10,000 years During repeated glacial cycles Possibility of an initial penetrating defect in one or a few canisters. Higher flow rate or lower transport resistance than the target values for a few deposition holes. Groundwater composition outside the target range for a short time during operation and soon after closure for a few deposition holes. Low density areas in the backfill where sulphate reduction to sulphide cannot be ruled out. Erosion of buffer in some deposition holes due to long-term infiltration of meteoric water or dilute glacial meltwater. Canister failure by corrosion due to unfavourable groundwater conditions and buffer erosion. Canister failure due to shear displacements in fractures during ice-sheet retreat.

32 25 Assessment of radionuclide release scenarios Scenarios and cases In the Reference Case realisation (BS-RC) of the base scenario (BS) for radionuclide release, an incidental deviation is assumed whereby one canister with an initial penetrating defect of 1.0 mm diameter is emplaced in the repository. All other canisters are assumed to comply with quality requirements. The single defective canister is cautiously assumed to be located in a deposition hole with relatively unfavourable hydrogeological characteristics. Except for the single defective canister, all other EBS performance requirements are assumed to be met and upheld during the evolution. Other cases within the base scenario consider alternative, cautiously selected positions of defective canisters within the repository and consequent different flow path characteristics, alternative near-field and geosphere speciation of radionuclides, and delayed establishment of the transport path through the canister defect. Two variant scenarios are identified that are considered plausible: an enlarging defect and partial degradation of buffer (VS1); and failure of an initially intact canister by corrosion following buffer erosion (VS2). Three disturbance scenarios are identified that are considered very unlikely: accelerated insert corrosion rate (AIC); rock shear (RS); rock shear followed by buffer erosion (RS-DIL). Various calculation cases for these scenarios are considered to illustrate parameter and model uncertainties. Combinations of these scenarios are also considered, both with each other and with the base scenario. Finally, complementary deterministic and probabilistic sensitivity analyses and Monte Carlo simulations are carried out to investigate further uncertainties, for example, considering the possibility of multiple canisters with initial penetrating defects, alternative groundwater types, and alternative assumptions and parameter values for radionuclide transport. Analysis of the base scenario Reference Case Analysis of the Reference Case shows that the highest rate of radionuclide release is of C-14, which peaks at around 4500 years and then declines due to radioactive decay. Other, longer-lived, radionuclides Cl-36, I-129 and Cs-135 contribute at early times and dominate beyond a few ten thousand years. The dominant migration path is from the buffer into fractures intersecting the deposition hole; migration paths in the EDZ of the deposition tunnel or in the tunnel backfill are less important. Figure 6 shows the near-field release and geosphere release rates for the base scenario Reference Case, normalised with respect to the radionuclide-specific constraint for the radioactive releases to the environment defined in STUK Guide YVL D.5. The figure indicates that during the dose criteria time window (up to 10,000 years) the normalised activity release is almost four orders of magnitude below the criterion of one as also given in Para 313 of YVL D.5; beyond a few tens of thousands of years the normalised activity release rate decreases to between five and six orders of magnitude below one. The limited role of the geosphere in attenuating the peak release rate is related to the cautious assumption that the defective canister in the Reference Case is located in a

33 26 Figure 6. Evolution of the near-field and geosphere release rates for the base scenario Reference Case, with the release rate for each radionuclide normalised with respect to the regulatory nuclide-specific constraints for radioactive releases to the environment. Regulatory geo-bio flux constraint denotes the constraint of 1 for the sum of the ratios between the nuclide specific activity releases and the respective constraints given in YVL D.5. deposition hole with relatively unfavourable hydrogeological characteristics. The complementary analyses show that, for most locations, the assumed canister defect would result in much lower C-14 peak release as most C-14 would decay during the transport in the geosphere. Cases investigating emplacement of a canister with an initial defect at other cautiously selected locations in the repository show little variation from the Reference Case results; this is to be expected as the near-field flows and geosphere transport parameters are similar also in these cases. Results from cases considering alternative speciation, allowing isotopes of silver, molybdenum and niobium to migrate in anionic form are also virtually indistinguishable from those of the Reference Case. The key results in the biosphere modelling are the projections of the development of the surface environment during the first 10,000 years and the potential radiological impacts on humans, plants and animals living in that environment. The present assessment calculates pathway-, radionuclide- and biosphere object-specific annual doses to a person and combines them into landscape dose. The annual landscape doses to each exposed individual in the population form the dose distribution, from which the annual dose to a representative person for the most exposed people and the average annual dose to other people are identified. The potential radiological impacts on plants and animals are estimated by calculation of absorbed dose rates. The results from analysing the Reference Case (BSA-RC) are briefly summarised below and in more detail in Biosphere Assessment.

34 27 Surface environment development The projection of the development of the terrain and ecosystems in the surface environment for the next 10,000 years is presented in Terrain and Ecosystems Development Modelling. Some results for the Reference Case are presented in Figure 7. Doses to humans The screening analysis performed on the geosphere releases in the repository Reference Case BS-RC results in five radionuclides being propagated all the way through the biosphere modelling chain in calculation case BSA-RC. These radionuclides are C-14, Cl-36, Mo-93, Ag-108m and I-129. The calculated annual doses to a representative person within the most exposed group (E most_exp ) and for other people (E other ) are presented in Figure 8. The structure of the time profile of the activity releases in Figure 6 and the resulting annual doses are rather similar. The more irregular shapes of the dose curves are mainly caused by the development of the surface environment, i.e., changes in the sizes of contaminated biosphere objects and their ecosystem types. The calculated dose maximum for the most exposed group is msv and occurs at about 3,000 years after closure; the corresponding dose maximum for other exposed people is msv and about 1,000 years earlier. These results are about 6 to 7 orders of magnitudes below the regulatory radiation dose constraints. The results also show that the contribution from C-14 dominates the annual doses. Doses to plants and animals The absorbed dose rates for plants and animals for the calculation case (BSA-RC) are all below mikrogy/h; the highest dose rates are observed for Pike and Beaver in the freshwater environment. All results for calculated absorbed dose rates to plants and animals are reported in Biosphere Assessment. Analysis of variant scenarios In Variant Scenario 1 (VS1) it is assumed that processes occurring at the buffer/rock interface lead to degradation of the outer part of the buffer and partial loss of its radionuclide retention capacity. Furthermore, there is an initial penetrating defect in one of the canisters. Enhanced transport of corrosive agents, such as sulphide, from the rock to the canister when the buffer is degraded may accelerate corrosion of the insert of this defective canister, as well as the overpack. It is assumed that the defect thus becomes enlarged over time due, for example, to volume expansion of the insert as it corrodes or to corrosion of the copper overpack. Results from cases designed to represent VS1 show that peak normalised release rates are about one order of magnitude higher than in the Reference Case, i.e. still almost three orders of magnitude below the regulatory requirement on the activity releases from the geosphere to the surface environment (regulatory geo-bio flux constraint). The peak is again dominated by C-14 but occurs later, at about 20,000 years, reflecting the influence of the progressively increasing diameter of the penetrating hole.

35 28 Figure 7. Surface environment projections for three time steps in the Reference Case. The approximate location of central part the repository is indicated with a red circle and the discharge locations with a green circle.

36 29 Annual dose [msv] 1.E-06 1.E-08 1.E-10 1.E-12 1.E-14 E_most_exp E_other BSA RC 1.E Year Figure 8. The annual doses to a representative person within the most exposed group (E_most_exp) and for other exposed people (E_other) for the calculation case BSA-RC. In Variant Scenario 2 (VS2), chemical erosion of the buffer takes place associated with ice-sheet retreat. Significant buffer erosion is considered unlikely, but cannot currently be excluded in at least some of the deposition holes. Eventually, advective conditions are established around the canisters in these deposition holes, leading to enhanced corrosion of the canister by sulphide, and eventually to canister failure (no initial penetrating defect is assumed but a thinner canister wall 35 mm is adopted, which is the minimum thickness according to the design specifications). Taking account of results of the modelling of buffer erosion and sulphide corrosion from the performance assessment, canister failure is not expected to occur for at least several hundred thousand years. At these long times, the geosphere release rate is dominated by the nonsorbing and long-lived radionuclides, namely I-129 and Cl-36. Modelled geosphere release rates also show periodic maxima, due to relatively rapid flushing of these radionuclides from the geosphere during periods of high flow associated with ice-sheet retreat. In the case representing the least favourable deposition position (VS2-H1), the peak normalised geosphere release rate is more than three orders of magnitude below the geo-bio flux constraint. The low peak normalised release rates calculated for a single failed canister in scenario VS2 indicate that VS2 indicate that the few canister failures that could potentially occur in the more likely lines of evolution (or even the few tens of canister failures calculated to occur based on highly pessimistic assumptions) could easily be tolerated without exceeding the regulatory constraint. The results of the calculation cases for the biosphere variant scenarios are reported in Biosphere Assessment. They show that the annual doses will not exceed the dose constraints set by the regulations.

37 30 Analysis of disturbance scenarios The rock shear (RS) scenario considers canister failure due to shear movements on fractures intersecting the deposition holes in the event of a large earthquake. Two cases have been analysed: RS1 and RS2, in which rock shear and canister failure are assumed to occur respectively at 40,000, i.e. during the present, temperate period, and at 155,000 years, during a period of ice-sheet retreat. The highest peak normalised release rates from the geosphere are, in both cases, more than two orders of magnitude below the regulatory geo-bio flux constraint. This implies that more than one hundred canisters would have to fail simultaneously before the regulatory geo-bio flux constraint would be exceeded, even without taking into account the low probability that this event would actually happen. This exceeds the few tens of canisters estimated to be in critical positions that are vulnerable to failure in the event of a large earthquake. In the scenario of rock shear followed by buffer erosion (RS-DIL), the buffer undergoes either immediate damage or longer-term erosion following the rock shear, due to the penetration of low-ionic strength water to repository depth. The peak release rates for RS-DIL cases are higher than for RS cases, but, nevertheless, the peak expectation value of the normalised release rate in the RS-DIL scenario is still around an order of magnitude below the regulatory limit. The accelerated insert corrosion scenario (AIC) considers the possibility that an initial penetrating defect in a canister becomes enlarged over time due to faster than expected corrosion of the insert whereas the performance targets are fulfilled for all the other engineered barriers and the host rock is expected to meet the target properties during the evolution for the whole time window. More pessimistically than in VS1, the enlargement of the defect is assumed to occur instantaneously at 15,000 years leading to complete loss of transport resistance of the defect. The analysis of this scenario focuses on the significance of whether or not a transport path between the canister interior and the buffer exists prior to defect enlargement. Two cases have been considered: AIC-TI assumes no path exists before enlargement and AIC-LI includes such a path. In both cases, release rates increase rapidly at 15,000 years to peak shortly thereafter. The peak is somewhat lower in AIC-TI compared with AIC-LI. The largest normalised releases from the geosphere are in both cases at least one order of magnitude below the regulatory constraint. Scenarios for inadvertent human intrusion caused by borehole drilling have been formulated and analysed. Expectation values of effective doses to drilling technicians and site geologists have been derived based on a stylised approach to the dose calculations and estimation of indicative annual probabilities of an intrusion event. The peak expectation value of the dose in the calculation case, where drilling affects the canister (DS(F)-HI-CANISTER) is around an order of magnitude below the regulatory radiation dose constraint for the most exposed people. The peak expectation value of the dose in the calculation cases where drilling affects contaminated buffer and backfill (DS(F)-HI-BUFFER and DS(F)-HI-BACKFILL) is several orders of magnitude below the regulatory radiation dose constraint for the most exposed people. Possible binary combinations of scenarios have also been considered. Many can be excluded from detailed analysis on qualitative grounds. Where it is appropriate to sum

38 31 the releases of two different scenarios, the combined normalised releases still do not exceed the regulatory constraint. The results of the calculation cases for other biosphere disturbance scenarios are reported in Biosphere Assessment. They show that the annual doses will not exceed the dose constraints set by the regulations. Complementary analysis Monte Carlo simulations and probabilistic sensitivity analysis (PSA) have been carried out for two model cases (Figure 9): 1. the hole forever case, where the initial penetrating defect in the canister overpack remains unchanged in size over time; and 2. the growing hole case, where the defect becomes instantaneously enlarged at a randomly sampled time. The PSA provides a rich source of understanding of the sensitivity of model outputs to variations in input parameter values, allowing the most important parameters and parameter combinations to be determined. It has been shown that C-14, Cl-36 and I-129 control the normalised release rates for both cases and that the peak of the mean release rates in the growing-hole case is about two orders of magnitude greater than in the holeforever case. The 10 % of the realisations with the greatest peak activity release rates to the surface environment are controlled by C-14 for both cases. The time to loss of hole resistance in the growing hole case, varied between 5000 and 50,000 years, is important in determining the peak release rate of C-14 (half-life 5700 years) whereas, for the longer-lived radionuclides, its influence is much smaller or even negligible. For longerlived, sorbing radionuclides, uncertainty in the diffusion and retention properties of the buffer are important in determining peak release rates.

39 32 1.E+00 Normalised release rate to the biosphere 1.E-01 1.E-02 1.E-03 1.E-04 1.E-05 1.E-06 1.E-07 1.E-08 1.E-09 Total Hole forever Maximum value 95 th percentile 99 th percentile Reference Case Mean Median 50 th percentile 1.E-10 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 Time (years) Regulatory geo-bio flux constraint 5 th percentile Minimum 1 st percentile value 1.E+00 Normalised release rate to the biosphere 1.E-01 1.E-02 1.E-03 1.E-04 1.E-05 1.E-06 1.E-07 1.E-08 1.E-09 Total Growing hole Maximum value 95 th percentile 99 th percentile Mean Median 50 th percentile 5 th percentile 1 st percentile Regulatory geo-bio flux constraint Minimum value 1.E-10 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 Time (years) Figure 9. Total normalised release rates to the environment in two Monte Carlo cases with 10,000 realisations (hole-forever and growing-hole cases). Summary of results and uncertainties The scenarios analysed address uncertainties in the evolution of the disposal system. A range of calculation cases has been analysed for each scenario. Case assumptions have been applied within each scenario that include cautious views on the severity of initiating events, subsequent degradation of engineered barriers and migration paths from defective or damaged canisters. Combinations of scenarios have also been

40 33 analysed. Parameter uncertainties have been investigated most thoroughly for the case of canister failure due to an initial penetrating defect; in this case deterministic analyses were complemented by Monte Carlo simulations and PSA. The model results are found to be consistent with the current understanding and show significant attenuation and delay of releases in the near field and geosphere. Figure 10 shows the peak normalised activity release rates from the geosphere to the surface environment for all calculation cases within the base, variant and disturbance scenarios. For cases RS1, RS2, RS1-DIL and RS2-DIL, 1000-year moving averaging has been applied before calculating the peaks, which is consistent with STUK Guide YVL D.5. The lowest peak normalised releases are for the Reference Case (BS-RC) and sensitivity cases within the base scenario. In all cases, peak normalised release rates to the surface environment are below the regulatory geo-bio flux constraint by around an order of magnitude or more. For the Reference Case the maximum of annual dose to a representative person within the most exposed group is about four to five orders of magnitude below the regulatory radiation dose constraint. Figure 10. Peak normalised geosphere release rates for all calculation cases within the base, variant and disturbance scenarios, each assuming the failure of a single canister. Colours are used to group cases by scenario. * indicates that 1000 year averaging is applied, in these cases. The right hand subfigure shows ranges of values for the peak probability-weighted normalised release rates in the RS and RS-DIL scenarios. These ranges arise due to uncertainties in the numbers of canisters failing due to rock shear, as well as the timing of failure.

41 34 Complementary considerations Complementary considerations provide additional evidence for the long-term safety of disposal according the KBS-3 method at the Olkiluoto site. Choice of geological disposal The choice of geological disposal as a concept for disposal of radioactive waste is backed by technical experience and international consensus. An appropriately chosen geological formation provides an environment that is stable over many millions of years geological timescales and the nature of changes that can occur is predictable from the geological sciences. A repository concept is developed that is consistent with the chosen geological formation, taking advantage of the benign or beneficial qualities and designed to withstand expected and unlikely events and processes that could affect the geological formation in the long term. The depth below ground provides buffering of the repository system from processes occurring in the surface environment and protection from unauthorised or inadvertent human intrusion. Support for the robustness of the KBS-3 method The KBS-3 method uses a few simple, common materials copper and iron for the canister, natural swelling clay for the buffer and backfill. This reduces the number of materials whose properties need to be understood and the number of interactions between the materials. The shorter-term properties of the materials are well known from their long use in engineering and industrial production in the case of copper and iron, and from their use in ground and underground engineering in the case of swelling clays. There is no difficulty in manufacturing or refining of the materials to the grade required by the KBS-3 design. Their longer-term properties are indicated from a range of natural analogues. Copper is one of the few metallic elements to occur in elemental form as a natural mineral native copper. There is evidence from a range of occurrences of native copper for very low corrosion rates of copper for millions of years under reducing conditions. Archaeological artefacts, while representing more variable and often more severe conditions, suggest low copper corrosion rates and also that localised corrosion is low compared with general corrosion. The bentonite buffer needs to maintain its low permeability and plasticity, and limit microbial activity. Studies of naturally-occurring bentonite deposits show that mineral alteration processes that are detrimental to the properties of low permeability and plasticity only occur significantly above about C even over geological timescales. Significant changes also depend on a supply of potassium, which will be limited in the buffer (due to a favourable groundwater composition and limited amount of potassium introduced to the system as foreign materials). Thus the bentonite buffer will remain stable during the repository thermal period, in which a maximum buffer temperature of around C is foreseen. There are several excellent examples of

42 35 bentonite and other clays functioning as a hydraulic barrier to preserve wood and human cadavers; these analogues also indicate that microbial activity was significantly reduced. Furthermore, radionuclide concentrations in the buffer and backfill have been shown to be comparable to examples of naturally-occurring radioactive material (NORM), and radionuclide release rates to be be generally comparable to, or less than, naturally occurring activity fluxes in groundwater at the site. Support for the suitability of the Olkiluoto site In Finland, there are limited choices of deep geological settings, leaving fractured crystalline basement as the only realistic choice of repository host rock. Such rocks have also been considered as suitable for locating a deep geological repository in many other countries, including several in which alternative rock formations are available. Key features of the Olkiluoto site include the stable tectonic situation, presence of suitable volumes of good quality rock suitable for repository construction, low groundwater flows, reducing conditions and also otherwise favourable groundwater conditions at repository depth. No natural resources have been found at Olkiluoto or nearby, reducing the risk of human intrusion. The Olkiluoto site is situated within the Fennoscandian Shield, away from active plate margins. In general, the frequency and magnitude of earthquakes in Finland is very low; earthquake magnitudes have never exceeded 5 (M=~5) since records began in the 1880s. Further, according to the data from historical earthquakes, the Olkiluoto area is located within a zone of lower seismicity, between two seismically active belts. There have been only nine recorded earthquakes within 100 km, with the nearest event (M=3.1) at km from Olkiluoto in An important consideration is to find sufficient volume of rock, with generally low and minor fracturing, to accommodate the spatial extent of the repository. Deposition tunnels must be placed avoiding shear zones or heavily fractured zones, although the access tunnel or shafts may cross such zones. At Olkiluoto, several options have been considered and suitable volumes of rock have been defined such that the deposition tunnels can be placed on a single level at a depth of m. The rock at Olkiluoto is geotechnically suitable for the construction of generally selfsupporting tunnels requiring only light rock support. Water inflows at depth are low and zones of inflow can be treated by local grouting. Significant local experience exists from construction of the ONKALO at Olkiluoto. Evidence from boreholes and the ONKALO show that, in the natural situation, groundwaters at repository depth are reducing and also have otherwise favourable hydrochemistry for the longevity and function of the canister, buffer and backfill. Levels of sulphide, which is expected to be the main agent for canister corrosion, and sulphate that might be reduced to sulphide at the repository depth, are both suitably low. There is no evidence of that dilute meltwater ever reached repository depth during past glacial cycles.

43 36 Conclusions The main research and development needs during the coming years The TURVA-2012 safety case assesses the performance and long-term safety of a KBS-3 type spent nuclear fuel disposal facility at Olkiluoto. The safety case also addresses the known uncertainties that may have an impact on the performance of the facility. The TURVA-2012 safety case forms the basis for the construction licence application in which Posiva proposes that the construction of the repository can be started. Some uncertainties still remain, but these do not affect the conclusions on longterm safety. Additional research and development will, however, help increase the reliability of the safety case to be compiled for the operational license application. The focus of the research and development in the coming years are on the: better understanding of the processes affecting canister corrosion and erosion of buffer and backfill; rock conditions in potential volumes of rock for the repository and the application of RSC criteria for the selection of repository panels, tunnels and deposition holes; demonstration of the implementation of the components of the repository system at full scale according to the technical design and quality performance requirements. Further investigations of the properties of the rock in the repository area will reduce the probability of locating the canisters in unfavourable positions with respect to future loads. The processes affecting the performance of the engineered barriers will continue to be experimentally studied. Technical tests will be applied to demonstrate that the repository can be implemented according to the assumptions made in the safety case. Conclusions on compliance The TURVA-2012 safety case has been compiled according to the regulatory requirements. It demonstrates that Posiva s repository design and analyses of performance and safety are fully consistent with all the legal and regulatory requirements related to long-term safety as set out in Government Decree 736/2008 and STUK YVL Guides. A detailed trace showing that each of legal and regulatory requirements is fulfilled is contained within the body of the TURVA-2012 portfolio. Key features of the demonstration are summarised below. The Posiva repository design is based on a robust system of multiple barriers. For the expected evolution of conditions in the Olkiluoto bedrock and engineered barriers, the copper canisters, in which the spent nuclear fuel is contained, are expected to contain all radionuclides for over one million years. The location of the repository, at a depth of about m below ground, will provide isolation from the surface environment and protection against inadvertent human intrusion. The mutually complementary barriers provide well-defined safety functions and the barriers are arranged so that the detrimental impact of a deficiency in any individual barrier on its safety functions will be compensated for by other safety functions. Similarly, the system of complementary barriers and safety functions provides robustness with respect to external events and processes, including geological and climatic changes. The requirements for the reliable operation of each safety function are

44 37 expressed in terms of performance targets for the engineered barriers and target properties for the host rock. These lead to design requirements for the engineered barriers and definition of a Rock Suitability Classification system (RSC), by which the local suitability of the rock for development of underground openings and deposition of spent nuclear fuel can be assessed. A comprehensive examination has been made of the features, events and processes that could affect the evolution of the disposal system (repository system plus surface environment), or the performance of individual barriers or fulfilment of their safety functions. Understanding of the changes due to construction and operation of the repository, and understanding of the longer-term natural processes (mainly related to climate changes) that will control the evolution of the natural setting of the repository, leads to the definition of future lines of evolution of the repository and its setting. The performance of the repository system has been systematically analysed in different time windows. The analyses take account of the uncertainties in the initial state and expected thermal, hydraulic, mechanical and chemical evolution of the repository system, and uncertainties in the expected future lines of evolution, and also the occurrence of unexpected or disruptive events. The analyses show that, under most conditions and lines of evolution of the host rock and engineered barriers, all performance requirements will be met. In this case, the copper canisters will remain intact and no releases of radionuclides will occur over at least one million years. Up to 50,000 years, the only plausible cause of release of radionuclides is that a canister with an initial penetrating defect has escaped detection and is emplaced in the repository. In the longer term, glacial episodes at the site may cause hydrogeological and hydrochemical changes leading to buffer erosion and increased canister corrosion and seismic disturbances leading to shear movements on fractures intersecting the deposition holes. These changes and disturbances, if they were to occur, could potentially lead to the failure of up to a few tens of canisters and to the release of radionuclides in less favourable locations within the repository. Although releases of radionuclides to the environment are not expected, the safety analyses focus on the cases in which releases of radionuclides could occur. It is shown that even accounting for unlikely combinations of emplacement of a canister with an initial penetrating defect in less favourable local rock conditions, peak normalised radionuclide release rates to the surface environment are orders of magnitude below the radionuclide-specific constraints specified in the STUK YVL Guide D.5. In the long term (approximately 100,000 years or more), calculated radionuclide release rates remain below the regulatory constraint for the radioactive release to the environment, even for pessimistic and unlikely combinations of damage to canisters by rock shear events and erosion of buffer material due to dilute groundwater conditions. Overall, it is concluded that the TURVA-2012 safety case demonstrates compliance with the legal and regulatory requirements for the planned and designed disposal facility for spent nuclear fuel at Olkiluoto. Some uncertainties still remain in the data and models, and some of these are unlikely to be eliminated. However, the analyses performed have shown that the repository system is robust against these uncertainties, and that the conclusions drawn about the compliance with the safety requirements hold even when these uncertainties are taken into account.

45 38

46 1 TABLE OF CONTENTS ABSTRACT TIIVISTELMÄ EXECUTIVE SUMMARY ABBREVIATIONS AND DEFINITIONS... 5 FOREWORD INTRODUCTION Spent nuclear fuel management in Finland Nature and evolution of the hazards presented by spent nuclear fuel Radiation risks Hazards and protection Evolution of the hazard and implications for design Posiva s programme for spent nuclear fuel disposal The TURVA-2012 safety case What is a safety case? The TURVA-2012 safety case The TURVA-2012 portfolio Quality assurance Legal and regulatory context for the management of spent fuel International treaties and agreements Legal requirements Regulatory guidance Safety submissions in support of construction and operating licences Feedback from STUK on the Interim Summary Report Structure of this report METHODOLOGY The KBS-3 method and the Olkiluoto site The KBS-3 disposal method The Olkiluoto site Design methodology Safety principles, safety concept and safety functions Performance targets and target properties Design requirements, rock suitability classification and design specifications Assessment methodology Iterative approach Description of the disposal system Features, events and processes Models and data and their use Assessment of performance of the repository system under the most likely lines of evolution Scenario formulation Approach to the analysis of radionuclide releases, transport and radiological impact Treatment of uncertainty Complementary considerations and supporting evidence Uncertainty management Quality management... 60

47 Goals and principles Application to TURVA-2012 safety case production Model qualification and code verification Data clearance Report and product review and approval process DESCRIPTION OF THE DISPOSAL SYSTEM Host rock Surface environment Underground openings and repository layout Spent nuclear fuel Canister Buffer Backfill and plug Deposition tunnel backfill Deposition tunnel plug Closure FEATURES, EVENTS AND PROCESSES Identification and screening of FEPs Identification of potentially relevant FEPs Screening for relevance to TURVA Organisation of the FEPs Development of the FEP descriptions FEP descriptions Coupling between FEPs and aggregation/disaggregation Onward use of the FEP descriptions Future lines of evolution MODELS AND DATA Models and data for climate evolution and climate-driven processes Key models and data for performance assessment and for formulation of radionuclide release scenarios Models and data for geosphere evolution Models and data for engineered barrier system performance Models and data for the analysis of radionuclide release scenarios Models and data for the biosphere assessment Development of surface environment Screening analysis Landscape modelling Radiological impact analysis PERFORMANCE ASSESSMENT OF THE REPOSITORY SYSTEM Excavation and operation up to closure of the disposal facility Repository system evolution and performance Fulfilment of performance targets and target properties Post-closure evolution during the next 10,000 years Repository system evolution and performance Fulfilment of performance targets and target properties Beyond 10,000 years during repeated glacial cycles Repository system evolution and evaluation of performance Fulfilment of performance targets and target properties Summary statement of performance and uncertainties

48 3 7 FORMULATION OF RADIONUCLIDE RELEASE SCENARIOS AND CALCULATION CASES Lines of evolution framing the scenarios and scenario formulation The link to scenario hierarchy Process for identification of scenarios and cases Repository system scenarios Base scenario for the repository system Variant scenarios for the repository system Disturbance scenarios for the repository system Radionuclide release scenarios and cases Surface environment scenarios Base scenario for the surface environment Variant scenarios for the surface environment Disturbance scenarios for the surface environment Calculation cases Summary and discussion on comprehensiveness Demonstrating that the set of scenarios is comprehensive Combinations of repository system scenarios ASSESSMENT OF RADIONUCLIDE RELEASE SCENARIOS Analysis of the Reference Case in the base scenario Results for the repository system Results for the surface environment Analysis of other cases in the base scenario Alternative canister positions BS-LOC1 and BS-LOC Alternative speciation BS-ANNFF / BSA-ANNFF Delayed establishment of the transport path BS-TIME / BSA- TIME Analysis of the variant scenarios in the repository system Cases in Variant Scenario 1 (VS1) Cases in Variant Scenario 2 (VS2) Analysis of the variant scenarios in the surface environment Analysis of the disturbance scenarios in the repository system Cases in the accelerated iron insert corrosion AIC scenario Cases in the rock shear RS scenario Case for the rock shear followed by buffer erosion in the RS- DIL scenario Analysis of the disturbance scenarios in the surface environment Cases in the human intrusion scenario (DS(F)-HI) Complementary analyses More than one defective canister in the repository Monte Carlo analyses and probabilistic sensitivity analysis Combinations of repository radionuclide release scenarios Summary of safety assessment results and uncertainties Geosphere release rates Doses to humans, animals and plants COMPLEMENTARY CONSIDERATIONS AND SUPPORTING EVIDENCE Choice of geological disposal Support for the robustness of the KBS-3 method Support for the suitability of geological disposal at the Olkiluoto site Safety and complementary indicators

49 4 10 COMPLIANCE WITH LEGAL REQUIREMENTS AND REGULATIONS AND ASSOCIATED UNCERTAINTIES Compliance with legal and regulatory requirements The main research and development needs during the coming years STATEMENT OF CONFIDENCE REFERENCES APPENDIX 1: GOVERNMENT DECREE (736/2008) APPENDIX 2: AUDIT OF LEGAL AND REGULATORY REQUIREMENTS RELATED TO THE LONG-TERM SAFETY CASE APPENDIX 3: REPOSITORY SYSTEM COMPONENTS FEPS AND SCENARIOS. 277

50 5 ABBREVIATIONS AND DEFINITIONS 3DEC ABAQUS AD AIC ALARA ANNFF AP BBM BFZ BIOPROTA BP BS BSA BWR CDF CFM CLIMBER-2 CLIMBER-2-SICOPOLIS CODE BRIGHT ConnectFlow A rock mechanics code, a three-dimensional numerical program based on distinct element method for discontinuum modelling. The finite element code used for calculating the glacially induced stresses, which combined with a synthetic regional background stress model, is used for assessment of the fault stability. Anno Domini. Scenario abbreviation: Accelerated Insert Corrosion. As Low As Reasonably Achievable. Scenario abbreviation: ANions in the Near and Far Field. After Present. Barcelona Basic Model, a critical state model that reproduces the mechanical behaviour of unsaturated soils under different boundary conditions. Brittle Fault Zone. Project which was set up to address the key uncertainties in long term assessments of contaminant releases into the environment arising from radioactive waste disposal ( Before Present. Base Scenario. Biosphere Assessment. Boiling Water Reactor (Olkiluoto 1&2). Cumulative Distribution Function. Colloid Formation and Migration test at Grimsel. CLIMate-BiosphERe model, an Earth System Model of Intermediate Complexity (EMIC) used for simulating the climate evolution. CLIMate and BiosphERe and SImulation COde for POLythermal Ice Sheets, models used in the future climate modelling. COupled DEformation BRIne, Gas and Heat Transport, the finite element code used to model the thermohydraulic behaviour of clay. The suite of groundwater modelling software that includes the NAMMU continuum porous medium (CPM) module as well as the NAPSAC discrete fracture network (DFN)

51 6 CPM DDM DFN DIL DiP Disposal facility module, which is used to develop DFN-based models groundwater flow and transport at the Olkiluoto site. Continuous Porous Medium. Displacement Discontinuity Method. Discrete Fracture Network (an approach used in groundwater flow modelling). Scenario abbreviation: DILute groundwater conditions, for example in RS-DIL, rock shear followed by buffer erosion. (Government) Decision-in-Principle. All underground tunnels, shafts, service areas and deposition panels (tunnels and holes), plus above-ground buildings that service the underground facility, but excluding the encapsulation plant. In this report, the above-ground parts are not discussed, as they are assumed to be dismantled upon closure and thereby have no effect on the long-term safety. Disposal system Repository system + surface environment. DP Dual Porosity groundwater modelling approach. DSn Scenario abbreviation: Disturbance Scenario. DZ-path Release path with exit from a deposition hole to the escavation damaged zone (EDZ) below the tunnel floor. EB Electron beam (weld/welding). EBS Engineered Barrier System, see Table 2-1. EBW Electron Beam Welding. Ecolego Simulation software tool used for creating dynamic models and performing deterministic and probabilistic simulations. ECPM EDZ Eh EMCL EMIC EPA EPR Equivalent Continuous Porous Medium. Excavation Damaged Zone; zone of the rock that is irreversibly damaged by the excavation of the tunnel. Redox potential. Environmental Media Concentration Limit. Earth system Model of Intermediate Complexity. U.S. Environmental Protection Agency. European Pressurised Water Reactor, trade name for the PWR reactor type for OL3.

52 7 ERICA FASTREACT FEFTRA FEM FEP F-path FPI Fracod2D FSAR FTRANS GAM GD GIS GoldSim GW HE HI HIPH Project under the EC 6 th Framework Programme aimed at providing an integrated approach to scientific, managerial and societal issues concerning the environmental effects of contaminants emitting ionising radiation, with emphasis on biota and ecosystems. FrAmework for Stochastic REACtive Transport, code used in hydrogeochemical modelling. The finite-element program package for groundwater flow modelling which applies the ECPM approach to model transient and density-driven flow and heat transfer by conduction and the DP approach for modelling salt transport. Finite Element Method. Feature, Event or Process (or as plural FEPs: Features, Events and Processes). Release path with exit from a deposition hole to a hostrock fracture intersecting the deposition hole. Full Perimeter Intersection, used to describe fracture extent in underground openings. A fracture mechanics code based on the Displacement Discontinuity Method (DDM) that has been used for predicting potential for spalling. Final Safety Analysis Report needed in support of an application for an operating licence. Code used in previous safety assessments for the analysis of the radionuclide release, retention and transport in the geosphere. Generalised Additive Model used to downscale nearsurface air temperature and precipitation from the CLIMBER-2-SICOPOLIS results. Government Decree. Geographical Information System. Code used for the analysis of the radionuclide release in the near field, and for probabilistic assessment in the near field and far field. Groundwater. HEterogeneous. Human Intrusion. Calculation case: HIghly alkaline (ph) water in geosphere and near field.

53 8 HIPH-NF Calculation case: HIghly alkaline (ph) water in the near field. Hn Canister position n in variant scenario V2. HZ IAEA ICRP IRF KBS KBS-3H KBS-3V K d L/ILW LDF LI Hydrogeological Zone. International Atomic Energy Agency. International Commission on Radiological Protection. Instant Release Fraction. (Kärnbränslesäkerhet). The method for deep geological disposal of spent nuclear fuel based on multiple barriers. (Kärnbränslesäkerhet 3-Horisontell). Design alternative of the KBS-3 method in which several spent nuclear fuel canisters are emplaced horizontally in each deposition drift. (Kärnbränslesäkerhet 3-Vertikal). The reference design alternative of the KBS-3 method, in which the spent nuclear fuel canisters are emplaced in individual vertical deposition holes. Distribution coefficient. Low and Intermediate Level (radioactive) Waste. Layout Determining Feature. LO1 2 Loviisa reactors 1 and 2. LOC LOT M MARFA MATLAB MPI/UW MX-80 NDT Calculation case: Leaky Insert, in AIC-LI calculation case. Calculation case: Canister location in calculation case BS- LOC. The long-term test of buffer material (at Äspö). Magnitude. Migration Analysis of Radionuclides in the FAr field: code, used to model radionuclide transport in geosphere. MATrix LABoratory (a numerical computing environment and programming language). Earth system model of Max Planck Institute used for the estimation of the climate evolution on a time scale of 10,000 years. Commercial name of the reference buffer bentonite. A high grade sodium bentonite from Wyoming, U.S., with a montmorillonite content of % (properties as specified in this report and references herein). Non-Destructive Testing.

54 9 NEA NORM OL1 2 OL3 OL4 ONKALO Pandora PDF PHREEQC POTTI PSA PSAR PWR QA QC Q DZ Q F QM Q TDZ RC RCC REPCOM Repository Repository system RIA RNT Nuclear Energy Agency. Naturally-Occurring Radioactive Material. Olkiluoto 1 and 2 reactors. Olkiluoto 3 reactor. Olkiluoto 4 reactor to be constructed at Olkiluoto. Expected to be similar to OL3 in TURVA-2012 safety case. Underground research facility constructed at Olkiluoto. Code used for radionuclide transport modelling in the biosphere. Probability Density Function. Reactive transport modelling code used for assessing the evolution of the groundwater chemistry. Database at Posiva. Probabilistic Sensitivity Analysis. Preliminary Safety Analysis Report a part of the construction licence application. Pressurised Water Reactor. Quality Assurance. Quality Coordinator. Flow rate in the DZ-path. Flow rate in the F-path. Quality Manager. Flow rate in the TDZ path. Calculation case: Reference Case in the Base Scenario (BS). Rank Correlation Coefficient. Code used in previous safety assessments for the analysis of the radionuclide retention and transport in the near field. Deposition tunnels + deposition holes. Spent nuclear fuel, canister, buffer, backfill (deposition tunnel backfill + deposition tunnel end plug), closure components and host rock. Excludes the surface environment. Radiological Impact Assessment. RadioNuclide release and Transport (Model/ling).

55 10 RQ RS RSC RTD RQ SAFCA SCC SFR SH SHYD SICOPOLIS Risk Quotient. Scenario abbreviation: Rock Shear (caused by an earthquake). Rock Suitability Classification system. Research, Technical Development and Design, see also TKS. Risk Quotient. The organisation of the TURVA-2012 safety case production process. Stress corrosion cracking. Sparsely Fractured Rock. Semi Homogeneous. Surface and near surface HYDrogeological model(ling). Ice-sheet model describing the evolution of the Northern Hemisphere ice sheets, their thickness and areal extent, basal temperature and bedrock elevation. SKB Swedish Nuclear Waste Management Company. SRB Sulphate Reducing Bacteria. SRRC Standardised Rank Regression Coefficient. SR-Site Safety assessment for a repository in Forsmark by SKB. STUK Finnish Radiation and Nuclear Safety Authority. TDS Total Dissolved Solids. TDZ-path Release path with exit from the deposition hole to the tunnel backfill above the deposition hole. TEM Ministry of Employment and the Economy, previously Ministry of Trade and Industry (KTM). TESM Terrain and EcoSystem development Model(ling). THM Thermal, Hydrological, Mechanical. THMC Thermal, Hydrological, Mechanical, Chemical. TI Calculation case: Tight Insert in calculation case AIC-TI. TILA-96 Name of Posiva s safety assessment TILA-99 Name of Posiva s safety assessment TKS Finnish equivalent for RTD (see RTD). TOUGHREACT An integral finite difference code used in thermo-hydrogeochemical modelling. TURVA-2012 Name of Posiva s safety case 2012, TURVA means safety.

56 11 TVO UNTAMO URL UVic VAHA VSn Variant Scenario n. VTT VVER-440 YEA YEL YJH YVL Teollisuuden Voima Oyj. Owner of the Olkiluoto power plants and co-owner of Posiva Oy. A GIS toolbox customised for Posiva for TESM. Underground Research Laboratory. Earth system model of the University of Victoria used for the estimation of the climate evolution on a time scale of 10,000 years. Requirements management system at Posiva. VTT Technical Research Centre of Finland. Pressurised water reactor type at Loviisa. Finnish abbreviation for Nuclear Energy Decree. Finnish abbreviation for Nuclear Energy Act. Finnish abbreviation for Nuclear Waste Management. STUK s (see STUK) regulatory guide series for nuclear facilities.

57 12

58 13 FOREWORD This report has been compiled and edited by Trevor Sumerling (Safety Assessment Management Ltd, UK) and Margit Snellman (Saanio & Riekkola Oy). Other contributors were Heini Laine, Nuria Marcos, Thomas Hjerpe, Pirjo Hellä and Annika Hagros (all Saanio & Riekkola Oy) and Paul Smith (SAM Switzerland GmbH).. The progress of the report was supervised by the SAFCA project group consisting of Ari Ikonen and Marja Vuorio (Posiva Oy), Pirjo Hellä, Thomas Hjerpe, Heini Laine, Nuria Marcos, Barbara Pastina and Margit Snellman (all Saanio & Riekkola Oy), and Paul Smith (SAM Switzerland GmbH). The report was reviewed at different stages by Juhani Vira (Posiva Oy), as well as the SAFCA project group and the various report contributors mentioned above. The final report review was carried out by the following individuals: Mike Thorne (Mike Thorne and Associates Limited, UK), Lawrence Johnson (Nagra, Switzerland), Ivars Neretnieks (KTH, Sweden), Allan Hedin (Executive Summary only) and Johan Andersson (SKB, Sweden), and Paul Degnan (Catalyst Geoscience - Geological & hydrogeological consultancy service, AUS). Their comments on the report are appreciated

59 14

60 15 1 INTRODUCTION This chapter introduces spent nuclear fuel management in Finland, the nature and evolution of the hazards presented by spent nuclear fuel, Posiva s programme for implementing its disposal at the Olkiluoto site, and the TURVA-2012 safety case that is required in support of Posiva s application for a licence to construct the disposal facility. The legal and regulatory context for the project, regulatory feedback from the previous safety submission, and the structure of this report are also set out. 1.1 Spent nuclear fuel management in Finland The spent nuclear fuel that arises from the generation of electricity at the Loviisa and Olkiluoto nuclear power plants is classified as nuclear waste. The safety and security of nuclear power plants, and the management of the wastes and spent nuclear fuel that result, are closely controlled within a framework of international treaties and agreements, and under national laws and regulations (see Section 1.5). In Finland, according to the Nuclear Energy Act of 1987 and including amendments made up to Act 410/2012: nuclear waste including spent nuclear fuel generated in Finland must be processed, stored and disposed of in Finland, and all practical and financial measures to ensure the safe and secure management and disposal are the responsibility of the nuclear power companies that produce the waste. This is consistent with both international treaties and conventions, and the ethical consensus for the management of such waste. In 1995, the two Finnish nuclear power companies, Teollisuuden Voima Oy (TVO) and Imatran Voima Oy (later Fortum Power and Heat Oy (Fortum)) established Posiva Oy (Posiva) to implement the final disposal programme for spent nuclear fuel and to carry out related research, technical design and development (RTD or TKS in Finnish). Other nuclear wastes are managed and disposed of by the power companies themselves. On assignment by its owners, Fortum and TVO, Posiva will take care of the disposal of spent fuel from the nuclear power plants at Loviisa and Olkiluoto. At Loviisa, two Russian-designed pressurised water reactors (VVER-440) are in operation; at Olkiluoto, two boiling water reactors (BWR) are operating and one pressurised water reactor (PWR) is under construction. Plans exist for a fourth nuclear power unit at Olkiluoto. At both sites there are facilities for interim storage of the spent fuel before disposal. In 2001, the Parliament of Finland endorsed a Decision-in-Principle (DiP) whereby the spent nuclear fuel produced by the operating Loviisa and Olkiluoto reactors will be disposed of in a geological repository at Olkiluoto. This first DiP allowed for the disposal of a maximum amount of spent nuclear fuel corresponding to 6500 tonnes of uranium (tu) initially loaded into the reactors. Subsequently, additional DiPs were issued in 2002 and 2010 allowing extension of the repository (up to 9000 tu) to accommodate spent fuel from the operations of the OL3 reactor and the planned OL4 reactor.

61 Nature and evolution of the hazards presented by spent nuclear fuel Radiation risks Exposure to radiation at high levels, or radioactive materials at high concentrations, can lead to detrimental health effects. Humans have been exposed to naturally occurring radiation and radioactive materials throughout their evolution, however, and it is established that at low levels of radiation exposure the risks are very small, such that the likelihood of any health detriment can only be estimated statistically, e.g. see NEA (1997). Two types of radiation health effect are distinguished: deterministic effects and stochastic effects. Deterministic effects are radiation effects the severity of which depends on the dose received and the effects are regarded as detrimental above some dose threshold; deterministic effects include tissue reactions, e.g. erythema, or organ damage. Stochastic effects are effects that are not certain to occur and the probability (but not the severity) of the effect is related to the radiation dose received and organs exposed 4 ; cancer is the primary concern. Systems of radiological protection, for workers and the public, aim to manage radiation exposures such that exposures are as low as reasonably achievable (ALARA) and in any case remain below defined constraints such that detrimental deterministic effects will not occur and that the probability of any stochastic health effect in any individual is very low. For example, the Government Decree 736/2008 requirement that the annual dose 5 to the most exposed people shall remain below the value of 0.1 msv (see Appendix 1) implies that the annual increment to the lifetime risk of death to the most exposed individual must be less than 10-5, i.e. less than one chance in 100, Hazards and protection Both humans and the environment have to be protected from the hazards presented by spent nuclear fuel. The two main radiological hazards are 6 : direct external radiation from the concentrated source as an intact unit, and external irradiation and also internal irradiation due to ingestion or inhalation of small amounts of the radioactive material (radionuclides) if it should become dispersed in the environment The exact relationship between dose and effect is uncertain especially at low doses. For radiation protection purposes, it is cautiously assumed that there is no dose threshold and the probability of a stochastic effect is proportional to the dose. In this report, annual dose refers to the sum of the effective dose arising from external radiation within the period of one year, and the committed effective dose from the intake of radioactive substances within the same year (GD 736/2008). Furthermore dose refers to effective dose, unless otherwise explicitly stated. The effective dose is the tissue-weighted sum of the equivalent doses in all specified tissues and organs of the body, where the tissue weighting factor represents the relative contribution of that tissue or organ to the total health detriment resulting from uniform irradiation of the body. The equivalent doses are mean absorbed doses in each tissue or organ, weighted by a factor that depends on the radiation type (ICRP 2007). Thus, effective dose is a quantity designed to reflect the amount of health detriment likely to result from the dose, based on current radiobiological, epidemiological and medical knowledge). A third hazard for spent nuclear fuel is the potential for nuclear criticality, which is prevented by design, see Description of the Disposal System, Section

62 17 Protection against these hazards is provided by isolation and containment. That is the spent nuclear fuel is placed: in a secure or inaccessible location away from humans and the environment, such that it cannot be reached except by deliberate actions, i.e. limiting accessibility; within a system of robust engineered and natural barriers such as to prevent any release of radionuclides, or to limit and attenuate any releases that might occur, i.e. preventing or limiting releases Evolution of the hazard and implications for design On its discharge from a nuclear reactor, spent nuclear fuel is highly radioactive, much more radioactive than any material found on Earth at the present-day. Simple point source calculations show that the external irradiation dose at 1 metre from 1 tonne of spent fuel at one year after discharge from the reactor is sufficient to deliver a fatal radiation dose within less than a minute (Hedin 1997). For this and other reasons spent nuclear fuel is very carefully managed such that access to the fuel is prevented by both physical barriers and procedural controls. At discharge from the reactor, spent fuel assemblies are placed in cooling ponds at the nuclear power plant sites. After a period of about 30 to 50 years, depending on the fuel type and its irradiation history, the radioactive heat output has reduced from an initial value of around 100 kw/tu shortly after at discharge to about 1 kw/tu. This level of heat output is low enough for encapsulation and disposal to proceed. The hundred-fold decrease in heat output results from the hundred-fold decrease in radioactivity taking place in this time. Figure 1-1 shows the activity of Finnish OL3 spent fuel relative to the activity of the uranium ore needed for its manufacture. The figure illustrates the relatively rapid decay of fission products and slower long-term decay of actinides such that the rate of decay declines after about 30 to 50 years. This indicates an appropriate time to move from storage to disposal, since the rate of decline of heat output and activity is slow by this time. Figure 1-1 also illustrates that after a few hundred thousand years the radioactivity of the spent fuel is similar to that of the uranium ore from which it was manufactured. This provides the basis for the design life-time of the canister in Posiva s design, which is set at hundreds of thousands of years. Design requirements for the other engineered barriers are set so as to support the canister life-time requirement, as discussed in Section

63 18 Figure 1-1. The total activity of one tonne of Finnish OL3 (denoted as EPR in the figure) spent nuclear fuel with a burn-up of 60 MWd/kgU is shown relative to the activity of 8 tonnes of natural uranium needed for its manufacture. The spent nuclear fuel activity is shown by type of radionuclide. 1.3 Posiva s programme for spent nuclear fuel disposal TVO began investigation work for the disposal of spent nuclear fuel in the late 1970s. Imatran Voima Oy (later Fortum) joined this work in the mid-1990s following the prohibition of the export of spent nuclear fuel from Finland. The programmes for location and development of a site for the disposal of spent nuclear fuel were united under Posiva Oy on its formation in Following extensive site investigations including national and regional surveys, and detailed investigations of five sites, Posiva selected Olkiluoto as the preferred site in A Government DiP endorsed by Parliament in 2001 affirmed that construction of a single disposal facility at Olkiluoto, serving the disposal needs of the four nuclear power units owned by TVO and Fortum, is in the overall best interests of society. Figure 1-2 provides a timeline for nuclear waste management for the Olkiluoto and Loviisa reactors in which the aim is to start the disposal of spent fuel around 2020.

64 19 Figure 1-2. Timeline for nuclear waste management relating to the Loviisa and Olkiluoto reactors until The target is to begin disposal of spent nuclear fuel around During the past few years, key activities in the programme have been related to: completion of the investigations for site confirmation at Olkiluoto both through analysis of data from surface-drilled characterisation holes and surveys and studies carried out in the ONKALO underground research facility, the design of the required surface and sub-surface facilities, the development of the selected disposal technology to the level required for the construction licence application, and demonstration of the long-term safety of the disposal of spent nuclear fuel including the preparation of a safety case presented as several separate reports, including the present report. Posiva s RTD (research, development and technical design) phase for the years was introduced in the TKS-2009 report (Posiva 2009a), which also provides insight into developments from previous RTD phases. The programme for (YJH-2012) was published earlier this year.

65 The TURVA-2012 safety case What is a safety case? Internationally, a safety case has been defined as a synthesis of evidence, analyses and arguments that quantify and substantiate the safety, and the level of expert confidence in the safety, of a geological disposal facility for radioactive waste (IAEA 2006, NEA 2004, 2012). The safety case is a key input to decision-making at several steps in the repository planning and implementation process. It becomes more comprehensive and rigorous as the programme progresses The TURVA-2012 safety case TURVA is Posiva s safety case in support of the Preliminary Safety Analysis Report (PSAR 2012) and application for a construction licence for a repository for disposal of spent nuclear fuel at the Olkiluoto site. It presents the long-term radiological safety case, which is concerned with the evolution, performance and safety of the disposal system following emplacement of the spent nuclear fuel. Other aspects of safety are dealt with in other parts of the PSAR. The TURVA-2012 safety case builds on an Interim Summary Report of the Safety Case 2009 that was published in 2010 (Posiva 2010a). The direction for further work to bring the safety case to maturity for submission in support of an application for a construction licence was laid out within Posiva s Review of Current Status and Future Plans for (TKS-2009, Posiva 2009a). TURVA-2012 is addressed to the nuclear regulator, STUK, and other national stakeholders as well as the international scientific and technical communities engaged in the discussion on nuclear waste disposal. STUK will review the safety case and related topical reports as part of its evaluation of the construction licence application and the PSAR. STUK will then give a statement on the construction licence application, which will form the basis for the Government judgement on issuance of the construction licence. The TURVA-2012 safety case presents the arguments for the long-term radiological safety of the planned disposal system. It includes: a description of the spent nuclear fuel to be disposed of in the geological repository a description of the natural and engineered barriers that the repository system provide, a definition of the safety functions and targets set for these and a description of the present understanding of the processes that may affect the evolution and performance of the repository system and the surface environment; a performance assessment systematically analysing the ability of the repository system to provide containment and isolation of the spent nuclear fuel for as long as it remains hazardous; a definition of the lines of evolution that may lead to failure of the canisters containing the spent nuclear fuel and to the releases of radionuclides (scenarios); 7 TURVA means safety in Finnish.

66 21 analyses of the potential rates of release of radionuclides from the failed canisters, the retention, transport and distribution of radionuclides within the repository system and surface environment and the potential radiation doses to humans, plants and animals including the associated uncertainties, and an evaluation of their impacts; the models and data used in the description of the evolution of the repository system and the development of the surface environment and for the analysis of activity releases and dose assessment; a range of qualitative evidence and arguments that complement and support the reliability of the results of the quantitative analyses; and a comparison of the outcome of the analyses with safety requirements. Aspects of safety related to the period of operations are dealt with in other parts of the PSAR. The safety case and supporting analyses will be further developed towards a Final Safety Analysis Report (FSAR) that will be submitted at the time of the operational licence application The TURVA-2012 portfolio The TURVA-2012 safety case is presented in a portfolio of safety case reports and supporting documents (Figure 1-3), and a synthesis that brings together all the lines of arguments for safety, including the main starting points, methodology, results and conclusions. The report names and brief descriptions of their contents are given in the figure. In this report, all TURVA-2012 portfolio reports are referenced using the report title (as in Figure 1-3) in italics. The full titles and report numbers are listed at the beginning of the reference list. The safety case portfolio has been developed based on the plan published in 2008 (Posiva 2008), which updated an earlier plan published in 2005 (Vieno & Ikonen 2005). Since 2008, the safety case and its presentation in the portfolio have been developed based on the feedback received from STUK. The updated safety case plan, which has directed the development of the TURVA-2012 safety case, places emphasis on quality assurance and control procedures and their documentation as well as on consistent handling of different types of uncertainties. Each safety case report has been subject to review in two steps: first, an internal review by safety case experts and other subject-matter experts within Posiva s RTD programme and, second, a review by external experts. Records of the external experts comments and Posiva s responses are documented and archived.

67 22 Biosphere Assessment: Modelling reports Description of the models and detailed modelling of surface environment Assessment of Radionuclide Release Scenarios for the Repository System TURVA-2012 Synthesis Description of the overall methodology of analysis, bringing together all the lines of arguments for safety, and the statement of confidence and the evaluation of compliance with long-term safety constraints Site Description Understanding of the present state and past evolution of the host rock Design Basis Performance targets and target properties for the repository system Production Lines Design, production and initial state of the EBS and the underground openings Description of the Disposal System Summary of the initial state of the repository system and present state of the surface environment Features, Events and Processes General description of features, events and processes affecting the disposal system Performance Assessment Analysis of the performance of the repository system and evaluation of the fulfillment of performance targets and target properties Formulation of Radionuclide Release Scenarios Description of climate evolution and definition of release scenarios Models and Data for the Repository System Models and data used in the performance assessment and in the analysis of the radionuclide release scenarios Data used in the biosphere assessment and summary of models Biosphere Assessment Analysis of releases and calculation of doses and activity fluxes. Complementary Considerations Supporting evidence incl. natural and anthropogenic analogues Main reports Main supporting documents Biosphere Description Understanding of the present state and evolution of the surface environment Biosphere Data Basis Figure 1-3. The TURVA-2012 safety case portfolio. The portfolio consists of safety case reports (green boxes) and supporting reports (blue boxes); brief descriptions of the contents are given (white boxes). Disposal system = repository system + surface environment.

68 Quality assurance The quality of the TURVA-2012 safety case has been assured through documented procedures in accordance with Posiva s quality management principle, which is based on the ISO 9001:2008 standard. A graded approach is applied whereby the primary emphasis is on quality control of those activities that have a direct bearing on long-term safety. Posiva s general quality guidelines are applied to the composition and quality management of portfolio reports and to the appointment of expert reviewers. Special attention is paid to the management of the processes that are applied to produce the safety case and its foundations. The purpose of this enhanced process control is to provide full traceability and transparency of the data, assumptions, models, calculations and results. The regulatory requirements on quality assurance are also followed. The overall plan, goals and constraints for the TURVA-2012 safety case production process are presented in Posiva s Safety Case Plan The organisation of the TURVA-2012 safety case production process is referred to as SAFCA. The details of how the Safety Case Plan is being implemented are described in the SAFCA project plan. The work is managed and coordinated by a SAFCA project group and supervised by a steering group. A SAFCA quality co-ordinator (QC) has been designated for activities related to quality assurance measures applied to the production of the safety case. Improvements are made to the process as deemed useful or necessary. The QC is also responsible for the coordination of the expert reviews, maintenance of schedules, and review and approval of the reports. Posiva s quality manager (QM) undertakes regular auditing of the safety case production process. Further details of the quality management system and its application to the production of the TURVA-2012 safety case are given in Section Legal and regulatory context for the management of spent fuel International treaties and agreements Finland is signatory to international conventions and treaties that define national obligations, standards of practice and protection, and reporting requirements for dealing with spent nuclear fuel. These include the: Joint Convention on the Safety of Spent Fuel Management and on the Safety of Radioactive Waste Management; Treaty on the Non-Proliferation of Nuclear Weapons; Convention on the Physical Protection of Nuclear Material; Treaties and Directives of the European Union.

69 24 Finland s international obligations for safe and secure management of nuclear materials and wastes, including spent nuclear fuel, on Finnish territory are translated into national laws. In particular the decision to seek permanent disposal of spent nuclear fuel in Finland is fully in accord with all relevant international treaties and agreements. The framework of legal and regulatory requirements to ensure safe and secure final disposal are as described in the following sub-sections Legal requirements The basis for the use of nuclear energy in Finland is given in the Nuclear Energy Act (YEL 990/1987) and Nuclear Energy Decree (YEA 161/1987), which came into effect in According to the Nuclear Energy Act: Nuclear waste shall be managed so that after disposal of the waste no radiation exposure is caused, which would exceed the level considered acceptable at the time the final disposal is implemented. and The disposal of nuclear waste in a manner intended as permanent shall be planned giving priority to safety and so that ensuring long-term safety does not require the surveillance of the final disposal site. According to the law, the Ministry of Employment and the Economy (TEM; previously the Ministry of Trade and Industry, KTM) decides on the principles to be followed in waste management of spent fuel and other nuclear waste. The safe management of nuclear waste is the responsibility of the utilities that generate the waste. The law also stipulates that the parties under the nuclear waste obligation must regularly submit to the Ministry a report setting out the responsible parties plans concerning the implementation of the measures associated with nuclear waste management and the preparation of these measures. Following the entry into force of the amendment to the Nuclear Energy Act in 2009, the reports must now be submitted every three years and include a description of the measures taken during the last threeyear period, as well as an outline of the plans for the next three years. The most recent report was submitted in The schedule for the disposal of spent nuclear fuel was first defined by the Government in 1983 and slightly modified by the Ministry of Trade and Industry (KTM) in 2003 (9/815/2003). According to the Ministry decision, the parties under the nuclear waste management obligation shall, separately, together or through Posiva Oy, present all reports and plans required to obtain a construction licence for a disposal facility for spent nuclear fuel by the end of The disposal facility is expected to become operational around The legislation concerning nuclear energy was updated in As part of the legislative reform, a number of the relevant Government Decisions were replaced with Government Decrees (GD). The Decrees entered into force on 1 st December The Government Decision (478/1999) regarding the safety of disposal of spent nuclear fuel, which particularly applied to the disposal facility, was replaced by Government Decree

70 25 736/2008, issued 27 November Government Decree 736/2008 sets the legal requirements regarding the safety of disposal of spent nuclear fuel. The scope and contents of Government Decree 736/2008 are summarised in Appendix Regulatory guidance The Radiation and Nuclear Safety Authority (STUK) issues guidance documents on the practical fulfilment of the legal requirements set out in Government Decree 736/2008. These guides also set out STUK s expectation for the content, quality and radiological criteria to be met by any safety case submission for disposal of nuclear waste in Finland. A total of five Guides apply to the disposal of spent nuclear fuel. The most relevant here are Guide YVL D.3, which provides guidance on the handling, storage and encapsulation of spent nuclear fuel and Guide YVL D.5 8, which provides guidance on the planning of the disposal method, design and operation of the disposal facility, safety requirements and demonstration of compliance with safety requirements, regulatory control and on the compilation of a safety case. Other Guides deal with nuclear nonproliferation control (Guide YVL D.1), transport of nuclear material and nuclear waste (Guide YVL D.2), and nuclear waste management and decommissioning activities (Guide YVL D.4). Guide YVL D.5 applies to disposal of all types of nuclear waste and provides guidance related to both operational and long-term safety. Key requirements, stemming from GD 736/2008 and set out in Guide YVL D.5, are summarised in Table 1-1. The Guide provides substantial additional information on the meaning of, and evidence needed to show compliance with, these requirements. The Guide YVL D.5 does not specify the precise time frames over which assessments are needed. Posiva consider, however, that radiation doses can be assessed, assuming human habits, nutritional needs and metabolism remain unchanged, with sufficient reliability over a period of up to 10,000 years, and that the fulfilment of the safety functions of the repository system and the release of radionuclide to the surface environment can be reasonably assessed up to one million years after repository closure. 8 The Guides YVL D.3 and YVL D.5 are available in draft form. STUK has agreed that the licence application can be based on version 4 of both Guides (version has been used).

71 26 Table 1-1. Synthesis of key requirements for long-term safety from STUK s Guide YVL D.5. Refer to the Guide for actual wording and context. Related to long-term radiological impacts For expected evolution scenarios, and in the period during which the radiation exposure can be assessed with sufficient reliability (at least over several millennia): the annual dose to the most exposed people shall remain below the value of 0.1 msv; the average annual doses to other people shall remain insignificantly low. In the longer term, the radiation impacts arising from disposal can at a maximum be equivalent to those arising from natural radioactive substances in Earth s crust, and on a large scale should remain insignificantly low. The nuclide-specific constraints on releases to the environment (average release of radioactive substances per annum) are specified in YVL D.5. For the activity releases that arise from the expected evolution scenarios, the sum of the ratios between the nuclide-specific activity release rates and the respective constraints given in YVL D.5 shall be less than one (evaluated for the release rates for radionuclides from the geosphere to the biosphere), The importance of unlikely events impairing long-term safety shall be assessed, and whenever practicable, the radiation impacts caused shall be assessed quantitatively. The resulting annual effective dose or activity release shall be calculated and multiplied by its estimated probability of occurrence. The obtained expectation value shall be below the dose constraint referred to above or release constraints given intable 2-4. The assessed radiation exposures to fauna and flora shall remain clearly below the levels that could cause decline in biodiversity or other significant detriment to any living population. Related to providing long-term safety Disposal shall be implemented in stages, with particular attention paid to aspects affecting long-term safety. The long-term safety of disposal shall be based on safety functions achieved through mutually complementary barriers so that a deficiency of an individual safety function or a predictable geological change will not jeopardise the long-term safety. Targets shall be specified for the performance of each safety function based on high quality scientific knowledge and expert judgement. For spent fuel, the safety functions provided by the engineered barriers shall limit effectively the release of radioactive substances into the bedrock for at least 10,000 years. The characteristics of the host rock shall be favourable for the long-term performance of engineered barriers and with respect to the groundwater flow regime at the disposal site Safety submissions in support of construction and operating licences According to the decision of the Ministry of Trade and Industry (KTM) in 2003 (see above), Posiva is to submit an application for a construction licence for a disposal facility at Olkiluoto by the end of This will be followed by an application to be made in 2018 for a licence to begin disposal operations. These applications will be accompanied by, respectively: a Preliminary Safety Analysis Report (PSAR) in support of the construction licence application and a Final Safety Analysis Report (FSAR) in support of the operating licence application. The PSAR and FSAR are to be prepared according to structures that are specified by STUK based on legislation and regulations. The PSAR and the FSAR will include the case for the operation of the disposal facility as a major nuclear facility and will cover both conventional and radiological safety during construction, operation, closure and in

72 27 the long term following completion of disposal operations and closure of the disposal facility. In particular, STUK specifies that the safety of the disposal facility in the long term should be assured through a safety case for disposal. 1.6 Feedback from STUK on the Interim Summary Report 2009 Following Posiva s submission of the Interim Summary Report of the Safety Case 2009 (Posiva 2010), the nuclear regulator, STUK, evaluated Posiva s preparedness to demonstrate long-term safety and operational safety, and the fulfilment of the safety requirements for nuclear waste disposal against the GD 736/2008. STUK noted that the material that Posiva submitted for the evaluation in 2009 covered the required elements at least on the design level, and that Posiva had done a considerable amount of research, development and design work preparing for nuclear waste disposal, and presented the acquired results in its reports. As a general observation, STUK noted that although Posiva presented a considerable amount of information in several safety-related areas, the information is not always consistently presented and sometimes it is difficult to trace. Therefore, in 2010, STUK was not able to reach a conclusion on the completeness of the material that Posiva proposed for the construction licence application. In addition, STUK considered that the material submitted had some shortcomings in demonstrating the fulfilment of the requirements and substantiating the conclusions drawn, because of limitations in the reasoning and analysis. STUK s safety evaluation report (STUK 2011) provided feedback and advice that has been taken into account as key issues that have been prioritised within Posiva s RTD programme and in the development of the TURVA-2012 safety case. The present safety case shows that the remaining shortcomings and uncertainties have an insignificant impact on long-term safety. The feedback has also been taken into account in the systematic structuring of the safety case and the reports included in the portfolio. The formulation of radionuclide release scenarios (see Formulation of Radionuclide Release Scenarios) follows a systematic approach taking into account the safety functions of the barriers of the repository system and the uncertainties in the features, events, and processes (see Features, Events and Processes) that may affect the disposal system from the emplacement of the first canister until the far future. Compliance with the performance targets and target properties (see Design Basis), which assures that the safety functions of the engineered and natural barriers will be achieved, is shown in Performance Assessment. This takes account of uncertainties in the initial state of the barriers and in the evolution of the repository system. 1.7 Structure of this report The present report is the TURVA-2012 Synthesis. It provides a description of the overall methodology of safety case, bringing together all the lines of argument, the evaluation of compliance with long-term safety constraints, and a statement of confidence in the TURVA-2012 safety case.

73 28 The structure of the report is as follows: Chapter 1 introduces spent nuclear fuel management in Finland, the nature and evolution of the hazards presented by spent nuclear fuel, Posiva s programme for implementing disposal at the Olkiluoto site and the TURVA-2012 safety case. The legal and regulatory context for the project, regulatory feedback from the previous safety submission, and the structure of this report are also set out. Chapter 2 introduces the KBS-3 disposal method and key features of the Olkiluoto site that bear on the development of the disposal concept and safety case for disposal of spent nuclear fuel at the site. The design methodology and assessment methodology are then set out. The Posiva quality management system, its application to production of the TURVA-2012 safety case, and model and data quality processes are also outlined. Chapter 3 presents a summary description of the disposal system in its initial state; that is, descriptions of the host rock and surface environment, and of the spent nuclear fuel and engineered barriers (canister, buffer, backfill and closure). These descriptions provide the basis for assessing the performance of the repository system and the safety of the disposal system. Chapter 4 describes the identification and screening of features, events and processes (FEPs) and development of a database of FEPs relevant to the performance assessment and analysis of potential radionuclide releases and radiological impacts. It also describes the onward use of the FEP descriptions in performance assessment and radiological modelling and analyses, and outlines the potential future lines of evolution of the repository system and surface environment. Chapter 5 describes the various models and data needed for the analyses supporting the safety case. The models are of four types: models describing the climate evolution and climate-driven processes; models to represent the FEPs that determine the evolution of the disposal system and that have to be taken into account to assess the performance of the engineered barriers and conditions in the host rock; models to analyse radionuclide release and transport from the near field through the geosphere to the surface environment; and models for biosphere assessment including models representing landscape development, radionuclide transport in the surface environment and potential doses or dose rates to humans, plants and animals. Chapter 6 summarises the performance of the repository system and demonstrates fulfilment of the performance targets and target properties for the engineered barriers and host rock. The performance assessment takes account of the expected thermal, hydraulic, mechanical and chemical (THMC) evolution of the repository system, and uncertainties in the expected evolution. The performance and fulfilment of performance requirements are considered for three periods: during excavation and operation up to closure; in the post-closure period during the next 10,000 years; beyond 10,000 years over repeated glacial cycles up to one million years. Chapter 7 summarises the formulation of radionuclide release scenarios and calculation cases. These focus on deviations in conditions and uncertainties in evolution, and unexpected events that could lead to the release of radionuclides. Scenarios are defined as a base scenario, variant scenarios and disturbance

74 29 scenarios; in relation to these a set of calculation cases (reference case, sensitivity cases, what if cases) is defined. Chapter 8 summarises the assessment of the radionuclide release scenarios and cases as defined in Chapter 7. This includes analysis of radionuclide release and transport in the repository system, simulation of the surface environment and analysis of potential radiological impacts on humans, plants and animals. Base, variant and disturbance scenarios are analysed, and uncertainties within these scenarios are investigated using a range of deterministic calculation cases as well as Monte Carlo simulations. Probabilistic sensitivity analyses have been carried out to assess sensitivities to parameter values and to explore the consequences of alternative model assumptions. Chapter 9 outlines complementary considerations that provide additional evidence for the long-term safety of disposal. Complementary considerations and additional evidence related to the choice of the geological disposal concept, the robustness of the KBS-3 method and the suitability of the Olkiluoto site. Selected results from evaluations of a range of complementary indicators for the repository system are also presented. Chapter 10 confirms the compliance with legal and regulatory requirements based on findings and results presented in the preceding chapters. It also outlines the main research and development needs during the coming years. Chapter 11 provides a statement of confidence, confirming that the TURVA-2012 safety case shows, at a level of detail appropriate to the repository construction licence application, that the safe disposal of spent nuclear fuel can be implemented through the KBS-3 method at the Olkiluoto site.

75 30

76 31 2 METHODOLOGY This chapter introduces first the KBS-3 disposal method and key features of the Olkiluoto site that bear on the development of the disposal concept and safety case for disposal of spent nuclear fuel at the site. The design and assessment methodology are then outlined. In the design methodology, the safety concept and safety functions are defined based on long-term safety principles. This leads to the development of the design basis, which includes the performance targets for the engineered barriers, target properties for host rock and design requirements for the repository system. The definition of the performance targets for the engineered barriers and target properties for the host rock, take into account the different loads and interactions, and features, events and processes (FEPs) that may act on the repository system at the time of canister emplacement and in the long-term. From the performance targets and target properties the design requirements are derived. Then, design specifications are worked out such that the fulfilment of these requirements can be verified during implementation. The performance of the repository system and its components is analysed taking into account the expected lines of evolution and the uncertainties involved. Conditions and events that could give rise to the release of radionuclides are identified, which provide the basis for the formulation of radionuclide release scenarios. These scenarios provide the basis for analyses of radionuclide release and transport, and of radiological impacts to humans, plants and animals. Complementary considerations and supporting evidence are also assembled in support of the performance and radiological safety analyses. 2.1 The KBS-3 method and the Olkiluoto site The 2001 DiP states that disposal of spent nuclear fuel shall take place in a geological repository at the Olkiluoto site, developed according to the KBS-3 method. The KBS-3 method and key features of the Olkiluoto site are outlined in the following subsections. Further information on the KBS-3V design and its technical realisation at the Olkiluoto site is given in Chapter The KBS-3 disposal method The KBS-3 method was conceived as a solution for the disposal of spent nuclear fuel in Sweden in the early 1980s. Since then, the method has been developed and its key elements tested by SKB in Sweden and Posiva in Finland, and in joint projects. The method envisages the disposal of spent nuclear fuel within a system of multiple barriers, which consists of engineered barriers and the natural barrier provided by the host rock. Posiva s reference design in the construction licence application is based on vertical emplacement of the spent nuclear fuel canisters individually in deposition holes (KBS- 3V) 9. 9 A potential alternative design of horizontal emplacement of multiple canisters in deposition drifts (KBS-3H) is being jointly developed by the Swedish Nuclear Fuel and Waste Management Company (SKB) and Posiva. The present safety case is based on the KBS-3V reference design.

77 32 In the reference design the repository is constructed on a single level with the floor of the deposition tunnels at a depth of between 400 and 450 m below the ground surface in the Olkiluoto bedrock (Figure 2-1). The spent nuclear fuel elements are placed into copper canisters with cast iron loadbearing inserts, and the canisters are emplaced vertically in individual deposition holes bored in the floors of the deposition tunnels. The canisters are surrounded by a swelling clay buffer material that separates them from the bedrock. The deposition tunnels, central tunnels, access tunnel, shafts and the other underground openings are backfilled with materials that help to restore the natural conditions in the bedrock after closure The Olkiluoto site The Olkiluoto site, located on the coast of south-western Finland (Figure 2-2), has been investigated as a potential site for disposal of spent nuclear fuel for over 25 years. This has included the construction of an underground rock characterisation facility the ONKALO. Olkiluoto Island has an area of about 10 km 2 ; the surface facilities including the encapsulation plant will occupy about 0.1 km 2 ; according to the current design and required capacity, the deposition tunnels and other tunnels will occupy about 2 km 2. The Olkiluoto site, as seen today, is the consequence of events and processes that have taken place over billions of years, from those reflected in the geological properties of the rocks forming the geosphere, to the much shorter-term changes related to more recent climate-driven processes: mainly changes in groundwater flow and groundwater composition and the geomechanical response to crustal movements related to glacial loading and unloading. A detailed description of the Olkiluoto site is given in Site Description, which describes the host rock, and in Biosphere Description, which describes the surface environment. Key features of the Olkiluoto site with respect to its suitability for the deep geological disposal of radioactive waste include: a stable tectonic situation within the Fennoscandian Shield, away from active plate margins; good quality crystalline bedrock suitable for the excavation of self-supporting tunnels and other underground openings, such as deposition holes, technical rooms and shafts; reducing conditions at disposal depths and also otherwise favourable geochemical characteristics of the groundwater, and low groundwater flow at depth occurs currently, as it has occurred over a long period in the past and it is expected to persist for a long period into the future.

78 33 Figure 2-1. Schematic illustration of the KBS-3V design at the Olkiluoto site. Figure 2-2. Olkiluoto Island is situated on the coast of the Baltic Sea in south-western Finland. Photograph by Helifoto Oy.

79 34 The conditions in the Olkiluoto bedrock provide favourable conditions for longevity and reliable functioning of the engineered barrier system (EBS). In addition, the low groundwater flows, and physical and chemical retardation processes, limit the movement of radionuclides. Key features and processes that provide constraints on the layout of the repository and other underground openings, or that must be taken into account in the assessment of long-term performance and safety include: presence of deformation and fractured zones, displaying more mixed geotechnical properties and, in some cases, increased hydraulic activity; higher rock stress at depth, which may cause disturbance to the rock making underground openings less stable; temperature and thermal conductivity of rock and residual heat output of the spent nuclear fuel; high salinity of groundwater at depth which may affect the performance of the engineered barrier system; continuing post-glacial crustal uplift and, in the longer term, climatic cooling and glaciation, leading to changes in rock stress and potential changes in groundwater flow and hydrochemistry, e.g. influx of dilute glacial melt waters into the host rock. Post-glacial crustal uplift and eustatic sea-level changes lead to relative changes in local sea level. Thus, a transition from a coastal to an inland environment is expected over the next millennia, which must be taken into account in the assessments of the potential impact of releases of radionuclides to the biosphere. 2.2 Design methodology Requirements management Posiva has developed a robust system design for geological disposal of spent nuclear fuel at Olkiluoto through a formal requirements management system (VAHA). This provides a rigorous, traceable method of translating the safety principles and the safety concept to a set of safety functions, performance requirements, design requirements and design specifications for the various barriers, i.e. a specification for realisation of the concept at the Olkiluoto site. The VAHA sets out: At Level 1, the stakeholder requirements that come from laws, decisions-inprinciple, regulatory requirements and other stakeholder requirements; At Level 2, the long-term safety principles, which lead to a definition of the safety concept and safety functions; At Level 3, the performance requirements consisting of performance targets for the engineered barriers, and target properties for the host rock, such that the required safety functions are fulfilled; At Level 4, the design requirements for the engineered barriers, and the underground openings, including rock suitability classification criteria (RSC criteria), such that the performance requirements will be met;

80 35 At Level 5, the design specifications, which are the detailed specifications to be used in the design, construction and manufacturing Safety principles, safety concept and safety functions Long-term safety principles The long-term safety principles set out for the KBS-3 method are based on the use of a multi-barrier disposal system consisting of engineered barriers and host rock. The role of the engineered barriers is to provide the primary containment against the release of radionuclides. The host rock should provide favourable conditions for the long-term performance of the engineered barriers, but also limit and retard the transport of radionuclides. The multi-barrier system as a whole should be able to protect the living environment even if one of the barriers turns out to be deficient. The long-term safety principles are described at Level 2 of the VAHA as follows. 1. The spent fuel elements are disposed of in a repository located deep in the Olkiluoto bedrock. The release of radionuclides is prevented with a multi-barrier disposal system consisting of a system of engineered barriers (EBS) and host rock such that the system effectively isolates the radionuclides from the living environment. 2. The engineered barrier system consists of a) canister to contain the radionuclides as long as these could cause significant harm to the environment b) buffer between the canisters and the host rock to protect the canisters as long as containment of radionuclides is needed c) deposition tunnel backfill and plugs to keep the buffer in place and help restore the natural conditions in the host rock d) the closure, i.e. the backfill and sealing structures to decouple the repository from the surface environment. 3. The host rock and depth of the repository are selected in such a way as to make it possible for the EBS to fulfill the functions of containment and isolation described above. 4. Should any of the canisters start to leak, the repository system as a whole will hinder or retard releases of radionuclides to the surface environment to the level required by the long-term safety criteria. The safety concept The safety concept (Figure 2-3) is a conceptual description of how these principles are applied together to achieve safe disposal of spent nuclear fuel in the present-day and future conditions of the Olkiluoto site. Containment of radionuclide inventory associated with the spent nuclear fuel is provided first and foremost by encapsulating the fuel in sealed (gas-tight and watertight) copper-iron canisters. The other EBS components (buffer, backfill and closure) provide favourable near-field conditions for the canisters to remain intact and, in the event of canister failure, slow down or limit releases of radionuclides from the canister. The containment of radionuclides is ensured by the proven technical quality of the EBS.

81 36 Other elements of the safety concept include sufficient depth for the repository, favourable and predictable bedrock and groundwater conditions and well-characterised material properties of both the bedrock and the EBS (the key safety features of the system in Figure 2-3). A robust system design ensures that single deficiencies in the design or implementation of the design, or uncertainties in future conditions, do not lead to significant weakening of the overall safe functioning of the repository system. Safety functions are assigned to the components of the engineered barrier system (EBS) and the host rock as shown in Table 2-1. The purpose of the multiple and complementary barriers, as described above, are to ensure that any single detrimental phenomenon or uncertainty cannot undermine the safety of the whole system, as required by Government Decree 736/2008: The long-term safety of disposal shall be based on redundant barriers so that deficiency in one of the barriers or a predictable geological change does not jeopardise long-term safety. SAFE DISPOSAL LONG-TERM ISOLATION AND CONTAINMENT FAVOURABLE NEAR-FIELD CONDITIONS FOR THE CANISTER Retention and retardation of radionuclides Slow transport in the geosphere Slow release from the spent fuel matrix Slow diffusive transport in the buffer PROVEN TECHNICAL QUALITY OF THE EBS FAVOURABLE, PREDICTABLE BEDROCK AND GROUNDWATER CONDITIONS SUFFICIENT DEPTH WELL-CHARACTERISED MATERIAL PROPERTIES ROBUST SYSTEM DESIGN Figure 2-3. Outline of the safety concept for a KBS-3 type repository for spent fuel in a crystalline bedrock (adapted from Posiva 2003). The safety concept is based on a robust system design. Orange pillars and blocks indicate the primary safety features and properties of the disposal system. Green pillars and blocks indicate the secondary safety features that may become important in the event of a radionuclide release from a canister.

82 37 Thus, for example, features and processes associated with the green columns in Figure 2-3 are, at least partly, independent of each other. The characterisation of the Olkiluoto site for the repository design is focused on a volume of bedrock situated between 400 and 500 metres below the ground surface. At such depths, the likelihood of inadvertent human intrusion is low, and favourable and predictable bedrock and groundwater conditions, such as reducing conditions, low frequency of water-conducting fractures and slow movement of groundwater, are found. The depth range is consistent with guidance in YVL Guide D.5 according to which the repository should be located:...at the depth of several hundreds of metres in order to mitigate adequately the impacts from aboveground natural phenomena, such as glaciation, and human actions. (YVL D.5, paragraph 412) Should any initially defective canisters be present or subsequent breaches in the canisters occur, the consequences of radionuclide releases for humans and other biota inhabiting the surface environment will be mitigated by the slow release from the spent nuclear fuel matrix, slow diffusive transport in the buffer and backfill, and slow radionuclide transport in the geosphere. Together, the engineered barriers and the rock provide for retention and retardation of radionuclides. Radioactive decay during transport also decreases activity releases into the environment. These are depicted in Figure 2-3 as secondary features of the safety concept (green blocks and pillars) since they become important only in the event of canister failure. Safety functions The long-term safety of disposal is based on a system of natural and engineered barriers which all have their roles in establishing the required long-term safety of the repository system. These roles constitute the safety functions of the barriers. According to YVL D.5, paragraph 405: Engineered barriers and their safety functions may consist of waste matrix, in which radioactive substances are incorporated; hermetic, corrosion resistant and mechanically strong container, in which the waste is enclosed; chemical environment around waste packages, which limits the dissolution and migration of radioactive substances; material around waste canisters (the buffer), which provides containment and yields to minor rock movements; other containment structures in the emplacement rooms; backfilling materials and sealing structures, which limit transport of radioactive substances through excavated rooms. Posiva s definition of safety functions follows this guidance with respect to engineered barriers (canister, buffer, backfill and closure). Most of the activity in the spent nuclear fuel is contained in a ceramic matrix (UO 2 ) that is resistant to dissolution in the expected repository conditions. The slow release of radionuclides from the spent fuel matrix in the event of canister failure is part of Posiva s safety concept. However, no safety functions or performance requirements are assigned to spent nuclear fuel; rather, the properties of the spent fuel are used as starting point in the design of the disposal system.

83 38 According to YVL Guide D.5, paragraph 406, the natural barriers and their safety functions may consist of stable and intact rock with low groundwater flow rate around disposal canisters rock around waste emplacement rooms where low groundwater flow, reducing and also otherwise favourable groundwater chemistry and retardation of dissolved substances in rock limit the mobility of radionuclides protection provided by the host rock against natural phenomena and human actions. In Posiva s repository concept, the natural barriers consist of the host rock, which carries the safety functions mentioned in the YVL Guide D.5, paragraph 406. The surface environment does not provide any safety functions; rather it is considered the object to be protected by the repository system. Thus, in summary, the disposal system consists of the surface environment, which is the object to be protected, plus the repository system, which consists of the spent nuclear fuel (the source of hazard) and the engineered and natural barriers that provide safety functions and thus protection from the hazard (Figure 2-4) The safety functions of the components of the engineered barrier system and host rock (barriers) as considered by Posiva are summarised in Table 2-1. Figure 2-4. The components of the disposal system. Safety functions are assigned to the natural barrier (host rock) and to components of the engineered barrier system (closure, backfill, buffer and canister).

84 39 Table 2-1. Safety functions assigned to the barriers (EBS components and host rock) in Posiva s KBS-3V repository. Barrier Canister Buffer Deposition tunnel backfill Host rock Closure Safety functions Ensure a prolonged period of containment of the spent fuel. This safety function rests first and foremost on the mechanical strength of the canister s cast iron insert and the corrosion resistance of the copper surrounding it. Contribute to mechanical, geochemical and hydrogeological conditions that are predictable and favourable to the canister. Protect canisters from external processes that could compromise the safety function of complete containment of the spent nuclear fuel and associated radionuclides. Limit and retard radionuclide releases in the event of canister failure. Contribute to favourable and predictable mechanical, geochemical and hydrogeological conditions for the buffer and canisters. Limit and retard radionuclide releases in the possible event of canister failure. Contribute to the mechanical stability of the rock adjacent to the deposition tunnels. Isolate the spent nuclear fuel repository from the surface environment and normal habitats for humans, plants and animals and limit the possibility of human intrusion, and isolate the repository from changing conditions at the ground surface. Provide favourable and predictable mechanical, geochemical and hydrogeological conditions for the engineered barriers. Limit the transport and retard the migration of harmful substances that could be released from the repository. Prevent the underground openings from compromising the long-term isolation of the repository from the surface environment and normal habitats for humans, plants and animals. Contribute to favourable and predictable geochemical and hydrogeological conditions for the other engineered barriers by preventing the formation of significant water conductive flow paths through the openings. Limit and retard inflow to and release of harmful substances from the repository Performance targets and target properties The safety functions described above are implemented in the proposed design through a set of technical design requirements, based on performance requirements that are defined for each barrier of the repository system. The performance requirements are expressed as performance targets (engineered barriers) and target properties (host rock) that the system should meet in the long-term to provide the required level of safety. In defining the performance targets for the engineered barriers, implementation aspects also have to be considered: the performance targets have to be set considering, on the one hand, the long-term safety aspects and, on the other hand, the need for the design and implementation to be robust as that is fundamental to the safety concept. The definition of the performance targets and target properties requires the identification of the different loads and interactions that may act on the repository system at the time of canister emplacement and in the long-term. To achieve this, the potential future conditions are described as alternative lines of evolution, and their likelihoods are assessed on the basis of present-day knowledge and the findings of earlier assessments. In the definition of the performance targets and target properties, all the lines of evolution and expected loads that are judged reasonably likely to occur (based on current understanding and previous findings) are taken into account and,

85 40 hence, included in the design basis. Performance assessment is used to show that the system, designed and built according to the design requirements and specifications, will meet the performance targets and target properties and thus that the safety functions will be fulfilled for all the reasonably likely lines of evolution. In this case there will be no radionuclide releases within the one million year time frame. Performance assessment also identifies any shortcomings that might occur in unlikely lines of evolution and these are carried forward to safety assessment wherein it is assessed whether the repository still will provide the protection level required by the regulations. This is consistent with STUK YVL Guide D.5, which states: Targets based on high quality scientific knowledge and expert judgement shall be specified for the performance of each safety function. In doing so, the potential changes and events affecting the disposal conditions during each assessment period shall be taken into account. In an assessment period extending up to several thousands of years, one can assume that the bedrock of the site remains in its current state, taking however account of the changes due to predictable processes, such as land uplift and those due to excavations and disposed waste. (STUK-YVL D.5, paragraph 408). In addition, the definition of the performance targets takes into account the requirement: The safety approach for disposal of spent fuel shall be that the safety functions provided by the engineered barriers will limit effectively the release of radioactive substances into bedrock for at least years. The derivation of the performance targets and target properties from the safety functions is described in Design Basis. Table 2-2 and Table 2-3, respectively, list the performance targets for the engineered components and target properties for the host rock, as catalogued in VAHA Level 3. These are derived based on the safety functions and expected evolution of the site. Table 2-2. Performance targets for (a) the canister, (b) the buffer, (c) the deposition tunnel backfill and plugs, and (d) closure. The performance targets, their rationale and the related design requirements are discussed in detail in the Design Basis. VAHA ID L3-CAN-4 L3-CAN-5 L3-CAN-7 L3-CAN-9 L3-CAN-11 L3-CAN-14 L3-CAN-16 L3-CAN-18 VAHA ID Performance targets a) Performance targets for the canister The canister shall initially be intact when leaving the encapsulation plant for disposal except for incidental deviations. In the expected repository conditions the canister shall remain intact for hundreds of thousands of years except for incidental deviations. The canister shall withstand corrosion in the expected repository conditions. The canister shall withstand the expected mechanical loads in the repository. The canister shall not impair the safety functions of other barriers. The canister shall be subcritical in all postulated operational and repository conditions including intrusion of water through a damaged canister wall. The canisters shall be stored, transferred and emplaced in such a way that the copper shell is not damaged. The design of the canister shall facilitate the retrievability of spent fuel assemblies from the repository. b) Performance targets for the buffer

86 41 VAHA ID L3-BUF-4 L3-BUF-10 L3-BUF-8 L3-BUF-12 L3-BUF-13 L3-BUF-14 L3-BUF-16 L3-BUF-17 L3-BUF-6 L3-BUF-19 L3-BUF-21 VAHA ID L3-BAC-5 L3-BAC-8 L3-BAC-9 L3-BAC-13 L3-BAC-16 L3-BAC-17 L3-BAC-18 L3-BAC-19 VAHA ID L3-CLO-13 L3-CLO-5 L3-CLO-6 L3-CLO-7 L3-CLO-8 L3-CLO-11 Performance targets Unless otherwise stated, the buffer shall fulfill the requirements listed below over hundreds of thousands of years in the expected repository conditions except for incidental deviations. The buffer shall mitigate the impact of rock shear on the canister. The buffer shall limit microbial activity. The buffer shall be impermeable enough to limit the transport of radionuclides from the canisters into the bedrock. The buffer shall be impermeable enough to limit the transport of corroding substances from the rock onto the canister surface. The buffer shall limit the transport of radiocolloids to the rock. The buffer shall provide support to the deposition hole walls to mitigate potential effects of rock damage. The buffer shall be able to keep the canister in the correct position (to prevent sinking and tilting). The buffer shall transfer the heat from the canister efficiently enough to keep the buffer temperature < 100 o C. The buffer shall allow gases to pass through it without causing damage to the repository system. The amount of substances in the buffer that could adversely affect the canister, backfill or rock shall be limited. c) Performance targets for the deposition tunnel backfill and plugs Unless otherwise stated, the backfill and plugs shall fulfill the performance targets listed below over hundreds of thousands of years in the expected repository conditions except for incidental deviations. The backfill shall limit advective flow along the deposition tunnels. The plugs shall isolate the deposition tunnels hydraulically during the operational phase of the repository. The chemical composition of the backfill and plugs shall not jeopardise the performance of the buffer, canister or bedrock. The backfill shall keep the buffer in place. The backfill shall contribute to the mechanical stability of the deposition tunnels. The plugs shall keep the backfill in place during the operational phase. The backfill shall contribute to prevent uplifting of the canister in the deposition hole. d) Performance targets for closure Unless otherwise stated, the closure materials and structures shall fulfill the performance targets listed below over hundreds of thousands of years in the expected repository conditions except for incidental deviations. Closure shall complete the isolation of the spent nuclear fuel by reducing the likelihood of unintentional human intrusion through the closed volumes. Closure shall restore the favourable, natural conditions of the bedrock as well as possible. Closure shall prevent the formation of preferential flow paths and transport routes between the ground surface and deposition tunnels/deposition holes. Closure shall not endanger the favourable conditions for the other parts of the EBS and the host rock. Retrieval of the spent nuclear fuel canisters shall be technically feasible in spite of repository tunnel and closure structures.

87 42 Table 2-3. Target properties for the host rock. The target properties and their rationale are discussed in detail in the Design Basis. VAHA ID L3-ROC-3 Target properties for the host rock Host rock shall, with the exception of incidental deviations, retain its favourable properties over hundreds of thousands of years. L3-ROC-5 The repository shall be located at minimum depth of 400 m. L3-ROC-10 L3-ROC-11 L3-ROC-12 L3-ROC-13 L3-ROC-14 L3-ROC-15 L3-ROC-16 L3-ROC-17 L3-ROC-29 L3-ROC-31 L3-ROC-19 L3-ROC-20 L3-ROC-21 L3-ROC-33 L3-ROC-23 To avoid canister corrosion, groundwater at the repository level shall be anoxic except during the initial period until the time when the oxygen entrapped in the near-field has been consumed. Therefore, no dissolved oxygen shall be present after the initially entrapped oxygen in the near-field has been consumed. Groundwater at the repository level shall a have high enough ph and a low enough chloride concentration to avoid chloride corrosion of the canisters. Therefore, ph shall be higher than 4 and chloride concentration [Cl - ] < 2M. Concentration of canister-corroding agents (HS -, NO - 2, NO - 3 and NH + 4, acetate) shall be limited in the groundwater at the repository level. Groundwater at the repository level shall have low organic matter, H 2 and S tot and methane contents to limit microbial activity, especially that of sulphate reducing bacteria. Groundwater at the repository level shall initially have sufficiently high ionic strength to reduce the likelihood of chemical erosion of the buffer or backfill. Therefore, total charge equivalent of cations Σq[M q+ ]*, shall initially be higher than 4 mm. * [M q+ ] = molar concentration of cations, q = charge number of ion. Groundwater at the repository level shall have limited salinity so that the buffer and backfill will maintain a high enough swelling pressure. Therefore, in the future expected conditions the groundwater salinity (TDS, total dissolved solids) at the repository level shall be less than 35 g/l TDS. During the initial transient caused by the construction activities salinities up to 70 g/l TDS can be accepted. The ph of the groundwater at the repository level shall be within a range where the buffer and backfill remain stable (no montmorillonite dissolution). Therefore, the ph shall be in the range of 5 10, but initially a higher ph (up to 11) is allowed locally. The acceptable level also depends on silica and calcium concentrations. Concentration of solutes that can have a detrimental effect on the stability of buffer and backfill (K +, Fe tot ) shall be limited in the groundwater at the repository level. Groundwater conditions shall be reducing in order to have a stable fuel matrix and low solubility of the radionuclides. In the vicinity of the deposition holes, natural groundwater shall have a low colloid and organic content to limit radionuclide transport. Under saturated conditions the groundwater flow in any fracture in the vicinity of a deposition hole shall be low to limit mass transfer to and from EBS. Therefore, the flow rate in such a fracture shall be in the order of one litre of flow per one metre of intercepting fracture width in a year (L/(m*year)) at the most. In case of more than one fracture, the sum of flow rates is applied. Flow conditions in the host rock shall contribute to high transport resistance. Therefore, migration paths in the vicinity of the deposition hole, shall have a transport resistance (WL/Q) higher than 10,000 years/m for most of the deposition holes and at least a few thousand years/m. Inflow of groundwater to deposition tunnels shall be limited to ensure the performance of the backfill. The properties of the host rock shall be favourable for matrix diffusion and sorption. The location of the deposition holes shall be selected so as to minimise the likelihood of the rock shear movements large enough to break the canister.

88 43 VAHA ID L3-ROC-30 Target properties for the host rock Therefore, the likelihood of a shear displacement exceeding 5 cm shall be low. To ascertain the data for sorption parameters, the ph shall be in the range of 6 10 after the initial period when a higher ph of up to 11 is allowed Design requirements, rock suitability classification and design specifications Design requirements and design specifications From the performance targets and target properties, the design requirements are derived. Further, the repository system is defined (by specifications) such that the fulfilment of the requirements can be verified at implementation. The long-term performance targets and target properties are used to derive the design requirements that the repository system must meet. For the engineered barriers these define the requirements that the barriers must meet in order to withstand the future expected loads. Design requirements form level 4 in VAHA. The performance targets, target properties and design requirements of each EBS component, and the underground openings and host rock are discussed in the Design Basis. Design specifications are the detailed specifications to be used in the design, construction and manufacturing that have been derived from the more general design requirements. They are defined so that the safety functions and performance targets are achieved initially and will be fulfilled in the expected conditions during the time that the spent nuclear fuel presents a significant hazard. Design specifications form level 5 in VAHA. The design specifications are discussed in the Canister-, Buffer-, Backfill, Closure- and Underground Openings Production Line reports for each component and summarised in Description of the Disposal System. Rock suitability classification For the rock barrier, the target properties set the starting point for the definition of the Rock Suitability Classification system (RSC) developed by Posiva. The Classification system includes both the updated rock suitability criteria as well the procedure for the suitability classification during the construction of the repository (McEwen et al. 2013). The RSC is used to identify suitable rock volumes for repository panels and to assess the suitability of deposition tunnels for locating deposition holes and to accept deposition holes for disposal. The aim is to avoid features of the host rock that may be detrimental to favourable conditions for safety either initially or in long term. The target properties presented in Table 2-3 outline the conditions that are considered to be favourable. The criteria developed for use in the classification system need to be based on observable and measurable properties of the host rock. These rock suitability

89 44 classification criteria (RSC criteria) constrain the rock properties around the repository. Based on interpretation, modelling and general understanding of the site properties, it is shown that the target properties for the host rock (Table 2-3) are fulfilled when the RSC criteria are met. Classification of the host rock according to RSC is carried out at different scales, including repository, panel, tunnel and deposition hole, and applied at different stages of the repository design and construction, proceeding from the layout design of the whole repository to the more detailed design and construction of panels, tunnels and, finally, deposition holes. The aims associated with the classifications at the different scales are as follows. Classification at the repository scale aims to define the rock volumes to be used for repository layout planning. Layout determining features (LDFs) are identified, as well as their respect volumes, which are to be avoided when locating deposition tunnels and holes. LDFs are either large fault zones that are potentially mechanically unstable in the current or future stress field, or they are main groundwater flow routes important or potentially important in the future for transport of solutes and chemical stability at the site. Classification at the panel scale aims to define suitable areas for the tunnels within a specific panel and to assess the degree of utilisation 10 of the panel area for the detailed design of the panel. The panel consists of a central tunnel and a number of deposition tunnels that will be excavated and used en bloc. The classification is done based on the more detailed data on deformation zones and hydraulically conductive zones that will become available during the construction of the central tunnels for the panel. The tunnel scale classification aims at defining suitable tunnel sections for the deposition holes, so that the LDFs and smaller, local deformation zones and their respect volumes, large fractures and high inflows to the deposition holes are avoided. At the deposition hole scale, the fulfilment of the rock suitability criteria is checked as part of the acceptance procedure for the individual deposition hole. The target properties and the rock suitability criteria are discussed further in Design Basis and in the Rock Suitability Classification report (McEwen et al. 2013). McEwen et al. (2013) also addresses the overall suitability and adequacy of the site as a natural barrier (YVL Guide D.5, paragraph 406), including checking of properties that would indicate unsuitability of the site, e.g. proximity of exploitable natural resources, abnormally high rock stresses with regards to the strength of rock (YVL Guide D.5, paragraph 410). 10 The degree of utilisation is determined by the number of suitable deposition holes with respect to the theoretical maximum number and is related to whether the volume of rock is being used in an economical and effective manner.

90 Assessment methodology Iterative approach The design basis (Section 2.2) has been developed, and performance and safety assessments (Section 2.3) have been carried out, in an iterative fashion. Figure 2-5 outlines the approach to the development of the safety case, whereby the design basis is developed, the performance of the repository system assessed, and scenarios leading to radionuclide release are formulated and assessed. The design basis and definition of performance targets and target properties are developed iteratively between performance assessment, formulation and assessment of radionuclide release scenarios and presentation of the safety case. Available scientific understanding, including the results from earlier assessments, is used in the definition of the performance targets, target properties for the host rock, design requirements and criteria for rock classification. These will be updated as scientific understanding is further developed, taking into account the results of the performance assessment and assessment of radionuclide release scenarios of the current safety case (the two-way arrows in Figure 2-5). The potential future conditions that are taken into account in the design process are described through a set of design basis scenarios. As required by regulation, the likelihood of different scenarios is assessed and those that are judged reasonably likely are included in the design basis scenarios. The performance targets and target properties, together with the derived design requirements and the underlying design basis scenarios, form the design basis of the repository. The design basis refers to the current and future environmentally induced loads and interactions that are taken into account in the design of the disposal system, and, ultimately, to the requirements that the planned disposal system must fulfil in order to achieve the objectives set for safety. A repository system designed and built according to the specified technical requirements will be compliant with the regulatory safety requirements. The situations in which the system does not fulfil the requirements, or there are significant uncertainties, or the evolution in the future is not to the design basis scenarios, are taken into account in the performance assessment and analysed in the safety assessment. The formulation and assessment of scenarios leading to radionuclide release is collectively termed safety assessment. In general, performance assessment and safety assessment provide feedback and guidance to the system design concerning: indications of the need for improved engineered solutions to increase robustness and confidence in the safety case; and specifications of the uncertainties and deviations that can be tolerated such that a performance target/target property is still achieved. The iteration between the design, performance assessment and safety assessment ensures, as far as possible: mutual compatibility of the engineered barriers with each other and with the bedrock, taking into account their respective safety functions;

91 46 resistance of the engineered barriers to the main thermal, hydraulic, mechanical and chemical loads to which they will be subjected during evolution of the system; and robustness with respect to slow processes and unlikely events that may occur over the regulatory compliance period, and a safety case that properly takes into account uncertainties in the implementation of the design (i.e. initial state uncertainties). Figure 2-5 outlines the approach to the development of the safety case, whereby the design basis is developed, the performance of the repository system is assessed, and scenarios leading to radionuclide release are formulated and assessed. Figure 2-5. Approach to the development of the safety case.

92 Description of the disposal system Posiva defines the disposal system as comprising the repository system (spent nuclear fuel, engineered barriers and host rock) and the surface environment (Figure 2-4). An accurate and reliable description of the disposal system is the foundation for development of understanding of possible lines of evolution, for assessments of performance and safety, and for development of complementary considerations that together comprise the safety case. Further, the STUK Guide YVL D.5 (draft) states: The preliminary and final safety analysis reports for a disposal facility shall include at least detailed description of the disposal site and description of its bedrock based on the investigations made so far description of the wastes to be disposed of, including their conditioning and packaging method and any materials to be installed around the disposed waste packages description of the disposal facility (excavations, engineered structures, and systems) and the way of its implementation (construction, operation and closure) Characterisation studies of the Olkiluoto site and bedrock have been made over a period of 25 years. This has lead to a detailed description and understanding of the site in respect of all characteristics relevant to its use for construction of a repository for spent nuclear fuel and in relation to its long-term evolution. Studies of the surface environment of the site form the basis for a description of the biosphere sufficient to characterise the environment to be protected and its potential future use and occupation by humans, plants and animals. Descriptions of the site and surface environment are provided in Site Description and Biosphere Description. The KBS-3 method and the KBS-3V design have been developed over more than 30 years. The specific realisation of the concept as planned for implementation of a repository for spent nuclear fuel at the Olkiluoto site is the result of thorough analysis of the functional requirements for the engineered barriers and host rock and of the overall safety of the repository system (as described in Design Basis). Detailed descriptions of the individual elements of the repository system and evidence concerning their practical realisation and feasibility are presented in Production Line reports (Canister, Buffer, Backfill, Closure and Underground Openings Production Line reports). Studies of spent nuclear fuel and its characteristics relevant to disposal have been carried out in many countries, and also in Finland with respect to the fuel types that will arise from the Olkiluoto and Loviisa reactors. The main characteristics and initial state, including uncertainties, of the repository system components (spent nuclear fuel, EBS and host rock) and of the surface environment to be used as an input to the safety assessment have been compiled in Description of the Disposal System. A summary description of the disposal system is given in Chapter 3 of this report.

93 Features, events and processes Identifying and describing the features, events and processes (FEPs) that are relevant to the evolution of the disposal system, or to its potential performance and safety, is an essential step towards ensuring completeness of the assessments and the safety case. For the TURVA-2012 safety case, the identification and description of FEPs has been carried out by a team of scientific subject and assessment experts, based on a review of the FEPs considered in previous assessments, the NEA FEP Database and FEPs considered in safety cases in other nuclear waste programme as well as examination of the specific characteristics of the Posiva disposal system and the Olkiluoto site. Following identification and description of FEPs, a systematic screening process is applied to rule out those FEPs that are not be relevant in the context of the repository system as proposed at the Olkiluoto site. The process took advantage of and drew on the experience from previous Posiva studies, as well as from the development of FEP lists in support of the assessment of the KBS-3V disposal method in Sweden. A FEP database has been developed providing a structured classification of relevant FEPs and couplings between them. In Features, Events and Processes relevant FEPs for the safety case are presented including a description of each FEP. Each description includes the current scientific understanding and relevance in the context of the Posiva disposal system at the Olkiluoto site, plus a note of any fundamental uncertainties in the scientific understanding. The FEP descriptions are organised according to the main components of the disposal system: Spent nuclear fuel; Canister; Buffer; Backfill; Auxiliary components 11 ; Geosphere; Surface environment; along with relevant External features, events and processes are discussed. For each component, FEPs affecting the physical state of the disposal system (evolution-related FEPs) and FEPs that mostly affect the transport of radionuclides (migration-related FEPs) are also distinguished. The process of developing the FEP database and the complete set of descriptions, references and other information is presented in Features, Events and Processes. The process is summarised and a list of retained FEPs is given in Chapter 4 of this report Models and data and their use Performance assessment and safety assessment require a range of models and input data. In some cases the models may be relatively simple, e.g. to determine the rate at which a single process may proceed for given conditions; in other cases, they may be complex models of coupled processes, that may vary in space and time, and/or take 11 Auxiliary components refer to backfilling of central tunnels, service areas, access tunnel and shafts, and seals and plugs that are installed both at the mouths of the deposition tunnels and as part of closure.

94 49 account of changing boundary conditions. In general, two types of model are distinguished: detailed models that aim at a realistic description of specific processes sometimes termed process models ; more simplified models used for scoping the impact of key processes and for analysing radionuclide release, retention and transport in a cautious manner. The first class of models are used to gain a realistic understanding of the possible evolution of aspects of the disposal system, e.g. the response of the groundwater flows and salinity to excavation of the repository or impacts of a glacial episode on the evolution of temperature and rock stress around the repository. These define the range of THMC conditions and loads under which the repository components must maintain their safety functions. More cautious, simple calculations may then suffice to show that safety functions are preserved. For example, even for pessimistic assumptions of groundwater composition and supply of oxidants, it is possible to use a simple calculation (see Appendix B in Performance Assessment) to show that the rate of corrosion of the copper canisters is so slow that canisters will not fail by corrosion of the copper canisters is so slow that the number of canisters failing by corrosion within the one million year assessment time frame remains limited. Another example of detailed modelling is geochemical modelling to support the evaluation of radionuclide speciation and retention parameters for use in assessments of radionuclide release cases. Input from hydrogeological and geochemical modelling may provide direct input to radionuclide release, retention and transport modelling, or, less directly, support judgements regarding input parameter value selection. In the case of biosphere modelling, detailed models of the evolution of the future landscape at the site provide the template for defining future ecological conditions and human uses of the environment that are taken into account in radionuclide transport modelling in the biosphere and in the radiological impact modelling. Radionuclide release and transport models, and also radiological impact models are defined and used in accord with STUK s Guide YVL D.5, which states: Simplifications of the models and the determination of the required input data shall be based on the principle that the performance of a safety function will not be overestimated while neither overly underestimated. (A06) and Selection of the computational methods, performance targets and input data shall be based on the principle that the actual radiation exposures or quantities of released radioactive substances shall with high degree of certainty be lower than those obtained through safety analyses. The uncertainties included in the safety analysis shall be assessed by means of appropriate methods, e.g. by sensitivity analyses or probabilistic methods. (A08) Model simplifications are applied where processes or data are uncertain, but must always be implemented in such a way that they can be seen to be cautious, e.g. by

95 50 reference to alternative models, by omission of processes that can only be beneficial to performance or safety, or by selecting input data such that impacts cannot be underestimated. Such simplifications may reduce data requirements, which is advantageous as it allows the effort of data gathering and quality assurance to focus on key parameters. Key input data for the performance assessment is the initial state of the repository system and the barriers provided by the Production Line reports summarised in Description of the Disposal System. Other key data reports are the canister design report (Raiko 2012), thermal dimensioning report (Ikonen & Raiko 2013) as well as the design of the disposal facility report (Saanio et al. 2013) presenting the repository layout. Site Description provides the description of the bedrock and the groundwater system, and the interacting processes and mechanisms. Biosphere Description provides the description of the present state and evolution of the surface environment. Several studies based on testing and modelling have been carried out to describe the evolution of the disposal system the evolving conditions at the disposal site and performance of the EBS under different conditions, e.g. groundwater flow, terrain and ecosystem development, buffer and backfill saturation, and erosion and canister corrosion. A summary of the results of these studies for the repository system is given in Performance Assessment and for the biosphere in Biosphere Assessment. These studies form the basis for the selection of the parameter values to be used in the safety assessment. For the repository system, the models together with the data forming the basis for the selection of the specific parameter values are described in Models and Data for the Repository System. As to the surface environment, the data used in the biosphere assessment are summarised in Biosphere Data Basis, and the models are discussed in Terrain and Ecosystems Development Modelling, Surface and Near-Surface Hydrological Modelling, Biosphere Radionuclide Transport and Dose Assessment and Dose Assessment for Plants and Animals. An overview of models and data is provided in Chapter 5 in this report Assessment of performance of the repository system under the most likely lines of evolution Posiva s safety concept is based on long-term isolation and containment, which is achieved through robust engineered barrier system design and favourable geological conditions at the repository site, as discussed in Section 2.1. The Government Decree 736/2008 (Section 11) states: The long-term safety of disposal shall be based on safety functions achieved through mutually complementary barriers so that a deficiency of an individual safety function or a predictable geological change will not jeopardise the long-term safety. Safety functions shall effectively prevent releases of disposed radioactive materials into the bedrock for a certain period, the length of which depends on the duration of the radioactivity in waste. For short-lived waste, this period shall be at least several hundreds of years, and for long-lived waste, at least several thousands of years.

96 51 The Guide YVL D.5 (draft 4, in Finnish only) further advises: Performance targets for the safety functions of engineered barriers shall be specified taking account of the activity level of waste and the half-lives of dominating radionuclides. The safety approach for disposal of spent fuel shall be that the safety functions provided by the engineered barriers will limit effectively the release of radioactive substances into bedrock for at least years. (Para 408) The aim of performance assessment is to present the evidence that the safety functions (as set out in Table 2-1) will be fulfilled, which being so, lead to isolation of the spent nuclear fuel, and complete containment over hundreds of thousands of years. Performance is assessed for the design basis scenarios, which define the envelope of future conditions taking into account the reasonably expected lines of evolution. These are also taken into account in the definition of the performance targets and target properties for the repository components. The performance of the repository system is analysed and the fulfilment of the performance targets and target properties (as set out in Table 2-2 and Table 2-3) is evaluated taking into account the expected thermal, hydraulic, mechanical and chemical (THMC) evolution of the repository system and the uncertainties in the expected lines of evolution. The possibilities for occurrence of less expected or disruptive events and processes, that could lead to reduction of one or more safety functions and, potentially, give rise to radionuclide releases, are also identified. Account is taken of the natural evolution of the environment, chiefly driven by climatic evolution, which imposes external loads on the repository system, and also internal loads, chiefly from the effects of excavation and emplacement of the spent nuclear fuel and the engineered barriers. Performance Assessment covers the performance of the system for the entire assessment time frame of one million years with a special focus on the containment safety function of the canister and isolating safety function of other EBS components and the geosphere in the first 10,000 years (as required by YVL D.5 paragraph 408). The performance is considered in three time windows: (1) during the excavation and operational period up to closure; (2) up until 10,000 years after closure; (3) beyond 10,000 years over repeated glacial cycles. The fulfilment of performance targets and target properties in each time window is assessed considering time-dependent and space-dependent loads on the engineered barriers and host rock. The assessment first considers the initial state of the repository system as defined in the production line reports (including deviations and uncertainties) constrained by the requirements and implemented according to design specifications. The performance targets and target properties (as defined in VAHA L3) are checked against evolutionrelated FEPs to ensure that the relevant processes and factors that could pose a threat to the performance of the barriers have been identified. The performance of the repository system and analyses of the response of the barriers to the evolution is then considered. The discussion is based on existing data and knowledge available in published reports and literature. Whenever possible, quantitative arguments are used, for example to provide estimates of safety margins and evidence of robustness of design.

97 52 Performance in each time window is summarised by a discussion on the state of the barriers with respect to the performance targets and target properties at the end of each of the periods, uncertainties are highlighted and the likelihood and effects of the deviations estimated. Special attention is given to identifying conditions and events (incidental deviations) that could lead to the release of radionuclides; these are taken forward to the formulation of radionuclide release scenarios, see Section The process is fully documented in Performance Assessment. A summary description of the main findings is given in Chapter 6 of this report Scenario formulation Uncertainties in the evolution and performance of the repository system and the surface environment mean that, in safety assessment, a range of scenarios, each representing one or more possible time histories of conditions, or lines of evolution, must be formulated and analysed. Consistent with the regulatory and international guidance (Guide YVL D.5; and IAEA 2009, 2011, 2012), Posiva distinguishes between the expected evolution of the disposal system and unlikely events and processes; account is also taken of the time window (or windows) in which releases of radionuclides might occur. Guide YVL D.5 states: Compliance with the requirements concerning long-term radiation safety, and the suitability of the disposal method and disposal site, shall be proven through a safety case that must analyse both expected evolution scenarios and unlikely events impairing long-term safety. The Guide goes on to define three types of scenarios: Base scenario: The base scenario shall assume the performance targets for each safety function, taking account of incidental deviations from the target values. Variant scenarios: The influence of declined performance of a single safety function or, in case of coupling between safety functions, the combined effects of declined performance of more than one function shall be analysed by means of variant scenarios. Disturbance scenarios: Disturbance scenarios shall be constructed for the analysis of unlikely events impairing long-term safety. The classification of scenarios is illustrated in Figure 2-6. The repository system is designed in a way that, for the design basis scenarios, except for incidental deviations, each component of the EBS meets the performance targets, assigned to it, and the host rock conforms to its target properties. In this case, the copper-iron canisters remain intact for the whole assessment time frame and there is no release of radionuclides. This is confirmed by the performance assessment. The performance assessment shows, however, that there are some plausible conditions and events (incidental deviations) that could lead to reduction of one or more safety

98 53 Figure 2-6. Classification of scenarios in TURVA-2012, which is consistent with STUK s Guide YVL D.5. functions, and thus may give rise to radionuclide releases. In addition, there are some very unlikely events and processes that could disrupt the repository, e.g. related to human intrusion and rock shear. These incidental deviations and unlikely events are systematically examined to define a set of scenarios that encompass the important combinations of initial conditions, natural evolution and disruptive events. In the current and past assessments by Posiva, the scenario of a canister with an initial penetrating defect has been considered in order to test the radiological performance of the other engineered barriers and host rock. This defect is most likely in the weld. Although the likelihood that a canister with an initial undetected penetrating defect will be emplaced is low, this is a useful base scenario for safety assessment (radionuclide release calculations) against which the efficiency of the other technical barriers and the host rock to limit the radionuclide releases can be tested and that also complies with the GD 736/2008. Thus, as indicated in Figure 2-6, the base scenario addresses the most likely lines of evolution (in which the performance targets and safety functions are met), but takes into account the possibility of one or a few canisters with initial undetected penetrating defects. Emplacement of a canister with an initial penetrating defect is not expected, but is an incidental deviation that cannot be ruled out. The variant scenarios address situations that are considered reasonably likely and in which there may be reduced performance of one or more safety functions of the barriers. Disturbance scenarios address the lines of evolution that are considered unlikely but cannot be completely eliminated. The process is documented in Formulation of Radionuclide Release Scenarios. A summary description of the scenarios is given in Chapter 7 of this report Approach to the analysis of radionuclide releases, transport and radiological impact The approach to analysis of radionuclide releases, transport and raiological impact, and in particular the endpoints to be calculated, are specified in legal and regulatory requirements. As discussed in Section 1.5, the Government Decree on the safety of disposal of nuclear waste (GD 736/2008) and the STUK Guide YVL D.5 set out the

99 54 requirements relating to radiological protection and provide guidance on the analysis of scenarios. Two time windows are distinguished in these documents, in which different protection criteria apply. For the earlier time window, which shall extend at a minimum over several millennia, Section 4 of GD 736/2008 states: The annual dose to the most exposed people shall remain below the value of 0.1 msv, and that The average annual doses to people shall remain insignificantly low. In the longer term, the quantitative regulatory criteria relate directly to radioactive releases to the environment. Guide YVL D.5 states: The sum of the ratios between the nuclide specific activity releases and the respective constraints shall be less than one and These activity releases can be averaged over 1000 years at the most. The nuclide-specific constraints referred to above are set out in Guide YVL D.5, paragraph 312. They form the basis for a sum-of-fractions approach to limiting radiological impact in the biosphere taking account of the relative potential for radiological impact of different groups of radionuclides. The probability of unlikely events giving rise to radionuclide releases may be taken into account when assessing compliance. Guide YVL D.5 states: The importance to safety of such unlikely events shall be assessed and whenever practicable, the resulting annual radiation dose or activity release shall be calculated and multiplied by its expected probability of occurrence. The obtained expectation value shall be below the radiation dose constraint or activity release constraint. Requirements for protection of species of fauna and flora are also set out, thus: Disposal shall not affect detrimentally to species of fauna and flora. This shall be demonstrated by assessing the typical radiation exposures of terrestrial and aquatic populations in the disposal site environment, assuming the present kind of living populations. The assessed exposures shall remain clearly below the levels which, on the basis of the best available scientific knowledge, would cause decline in biodiversity or other significant detriment to any living population. (Para 316) Consistent with regulatory guidance, the main safety indicators calculated and assessed in TURVA-2012 are: The radioactive releases from the bedrock to the biosphere (surface environment), which are calculated for all release scenarios and assessed against the nuclidespecific constraints for radioactive releases to the environment (average annual release rates of radioactive substances) defined in YVL D.5 (see Table 2-4); Annual doses to humans. Consistent with regulatory guidance these are calculated for scenarios that give rise to releases to the surface environment in the first 10,000 years. They are assessed against the requirements that:

100 55 The annual dose to the most exposed people shall remain below the value of 0.1 msv and The average annual doses to people shall remain insignificantly low. Absorbed dose rates to plants and animals. These are also only calculated for releases to the surface environment in the first 10,000 years. They are assessed against the requirement that the assessed exposures shall remain clearly below the levels that would cause decline in biodiversity or other significant detriment to any living population. The repository system is analysed using models that represent: release from the spent nuclear fuel (taking account of the locations of radionuclides in the fuel, its cladding and other parts of the fuel element); release, retention and transport in the near field (release from the canister, migration through the buffer, migration by alternative routes to water-conducting fractures in the host rock); and retention and transport in the geosphere (through water-conducting fractures taking account of variability in flow paths). This yields radioactive releases from the geosphere to the biosphere, which are used as input to biosphere models. Since the repository system models are run independently of, the biosphere models, the output from a single repository system calculation can be input to alternative biosphere models so as to represent alternative surface environment conditions at the time of release. In addition, complementary indicators are evaluated to broaden the understanding of the repository system performance with respect to retention and release of radionuclides. These indicators include total amounts of activity and activity concentrations in model compartments, activity fluxes between compartments and delay times (see Chapter 9). Table 2-4. Radionuclide-specific constraints for radioactive releases to the environment, as set out in STUK Guide YVL D.5. In a given row, the constraint applies to each individual radionuclide. The sum of the ratios between the nuclide specific activity releases and the respective constraints shall be less than one. These activity releases can be averaged over 1000 years at the most. The probability of unlikely events giving rise to activity releases may be taken into account. Radionuclides Constraints [GBq/a ] Long-lived alpha-emitting Ra, Th, Pa, Pu, Am and Cm isotopes 0.03 Se-79; Nb-94; I-129; Np C-14; Cl-36; Cs-135; long-lived uranium isotopes 0.3 Sn Tc-99; Mo-93* 3 Zr Ni Pd * Mo-93 is not mentioned in YVL D.5. However, based on a preliminary evaluation, STUK has recommended that the same nuclide-specific constraint as for Tc-99 is used.

101 56 Modelling for biosphere assessment includes, first, a screening process to identify those radionuclides that could make significant contributions to total radiological impacts. These radionuclides are carried forward to detailed biosphere modelling, based on a model of the future landscape and ecosystem development in the Olkiluoto region over the next 10,000 years. This provides the framework for modelling of radionuclide movements within compartments of the future surface environment and calculation of the radiation doses to humans, and to plants and animals, inhabiting or making use of the various areas and resources that may become contaminated. In TURVA-2012, to assess the performance of the repository system, the quantity presented is the normalised activity release rate of a given radionuclide, or the sum of normalised activity release rates. The normalised activity release rate is a dimensionless quantity defined as the activity release rate divided by the respective radionuclidespecific constraint, as given by Table 2-4. To satisfy the regulatory constraint on the release rates of activity from the geosphere to the biosphere (geo-bio fluxes), the normalised activity release rate, summed over all radionuclides, must be less than one Treatment of uncertainty Guide YVL D.5 (draft) requires: The uncertainties included in the safety analysis shall be assessed by means of appropriate methods, e.g., sensitivity analysis or probabilistic methods. The safety case shall include an assessment of the confidence level with regard to compliance with the safety criteria and of uncertainties with most contribution to the confidence level. Consistent with international best practice in safety assessments, uncertainties are analysed by a number of complementary methods, which include consideration of a range of calculation cases representing alternative future evolutions of the disposal system and the potential occurrence of unlikely events, alternative models of key processes, and uncertainties in data values. The derivation of scenarios that represent alternative future evolutions of the disposal system and the occurrence of unlikely events has been discussed in Section The remainder of this section considers the development of calculation cases to represent alternative scenarios, models and data. Calculation cases The radionuclide release scenarios (Section 2.3.6) form the framework for the definition of calculation cases that explore the uncertainties within each scenario. Calculation cases are defined to evaluate compliance of the repository with regulatory requirements on radiological protection, as well as to illustrate the impact of specific uncertainties or combinations of uncertainties on the calculated results. Each case illustrates different possibilities for how the repository might evolve and perform over time, taking into account uncertainties in the models and parameters used to represent radionuclide release, retention and transport, and radiological impact. While all uncertainties and combinations of uncertainties need to be considered in formulating the calculation cases, some combinations of uncertain model assumptions

102 57 and parameter values can be excluded on the grounds that they represent very unlikely or implausible outcomes. Thus a reduced set of calculation cases can be carried forward that, nevertheless, spans the domain of plausible outcomes. Uncertain model assumptions or parameter values can also be excluded if they can be argued, or shown, to yield lower consequences than existing calculation cases incorporating more cautiously chosen model assumptions or parameter values. Four types of calculation cases are distinguished: A Reference Case is one model realisation of the base scenario. Models and data for the Reference Case are, in most instances, selected to be either realistic or moderately cautious, i.e. radiological impacts are not to be underestimated nor excessively overestimated. Sensitivity cases represent alternate models and/or data to those of the Reference Case, but that remain within the scope of the base scenario and/or variant scenarios. Analyses of the sensitivity cases illustrate the effect of model and data uncertainties. What-if cases are mainly model representations of disturbance scenarios. Models and data for these what-if cases are selected to represent unlikely events and processes. Complementary cases are designed to develop a better understanding of the modelled system or subsystems. The number of calculation cases requiring biosphere modelling is reduced by considerations of conservatism and of likelihood, as discussed above. In particular, combinations of highly conservative/unlikely cases of the repository system model with highly conservative/what-if cases of the biosphere are avoided. All of the above cases are analysed deterministically, i.e. each calculation is carried out for a defined set of input parameter values. In addition, the disposal system behaviour is explored by Monte Carlo simulations and probabilistic sensitivity analysis (see below). Probabilistic sensitivity analysis (PSA) Many of the parameters used in the radionuclide release and transport calculations are affected by significant uncertainties, due to spatial variability, time evolution of the environmental conditions and statistical and systematic uncertainties in the parameter values. It is not practical, using individual, deterministically specified calculation cases, to explore the consequences of every possible combination of parameter values. Probabilistic sensitivity analysis, which complements the deterministic evaluation of calculation cases, is used to overcome this problem. The probabilistic sensitivity analyses (PSAs) carried out for TURVA-2012 were based on the technique of Monte Carlo simulation. In Monte Carlo simulation, performance measures, such as radionuclide release rates from the geosphere to the biosphere, are calculated a large number of times. Each calculation represents a different realisation of the modelled system, in which all uncertain input parameters are sampled randomly from probability density functions (PDFs). This means that single random values are selected from specified PDFs describing each parameter. As a result of this random

103 58 sampling from PDFs, each realisation is considered to be equally probable. For each realisation, all uncertain input parameters are sampled, meaning that a single random value is selected from a specified distribution describing each parameter (probability density function, or PDF). The results of the multiple realisations are assembled into probability distributions of possible outcomes. A statistical analysis is then carried out to identify, for example, the input parameters, or combinations of parameters (lumped parameters), for which variance in the input most affects the variance in the calculated performance measures. Different statistical analysis methods can be used. If the outcome is a similar ranking of important parameters, this shows the robustness of the sensitivity analysis. In TURVA-2012, the PSA has been carried out for the base scenario of the repository system, in which it is assumed that a canister with an initial penetrating defect is emplaced in the repository. The defective canister may be present at any location in the repository Complementary considerations and supporting evidence Complementary considerations provide additional evidence for the long-term safety of disposal according to the KBS-3 method at the Olkiluoto site. STUK s Guide YVL D.5 states: The importance to safety of such scenarios that cannot reasonably be assessed by means of quantitative safety analyses, shall be examined by means of complementary considerations. They may include e.g. analyses by simplified methods, comparisons with natural analogues or observations of the geological history of the disposal site. The significance of such considerations grows as the assessment period increases, and safety evaluations extending beyond a time horizon of one million years can mainly be based on the complementary considerations. Complementary considerations shall also be applied parallel to the actual safety assessment in order to enhance the confidence in results of the analysis or certain part of it. (A09) Complementary considerations and supporting evidence in support of the TURVA-2012 safety case have been assembled related to: the choice of geological disposal as a concept for disposal of radioactive waste, which is a choice backed by international accord; support for the robustness of the KBS-3 method, which makes use of a few simple materials with well-known properties established by engineering practice and by evidence from natural analogues; the favourable features of the Olkiluoto site, which include the stable tectonic situation, the presence of suitable volumes of good quality rock appropriate for repository construction, and low groundwater flows, reducing conditions and otherwise favourable groundwater conditions at repository depth. Arguments and evidence related to each of these points is compiled in Complementary Considerations; the arguments and evidence are summarised in Chapter 9 of this report. In addition, Chapter 9 summarises results from calculations of alternative indicators that

104 59 provide additional perspective on the level of protection provided by Posiva s KBS-3V repository system. 2.4 Uncertainty management Posiva has developed a systematic approach to the management of uncertainties in the safety case based on an iterative process where research, development, technical design and assessment results play an essential role (Figure 2-7). The overall strategy can be summarised as: identify, avoid, reduce and assess. Uncertainties need to be identified, i.e. described and quantified, and their relevance to safety needs to be considered. The safety case is largely composed of identification of uncertainties and assessment of their relevance. Uncertainties in the theoretical and conceptual understanding of FEPs are identified in Features, Events and Processes. The uncertainties in the model and data that describe the FEPs are reported in Models and Data for the Repository System and Biosphere Data Basis. Discussion of the representation and impact of uncertainties are discussed in Performance Assessment, Assessment of Radionuclide Release Scenarios for the Repository System and Biosphere Assessment, with respect to the uncertainties in each of these assessments. The development of the repository system is based on the idea of robustness, which means, where practicable, avoiding concepts and components the behaviour of which would be difficult to understand and predict. By means of research, it is possible to gain new knowledge and understanding of the system behaviour and, thereby, to reduce the uncertainties. The impact of uncertainties can also be reduced by technical design, for example by introducing safety margins. A practical example of reducing uncertainties is the development of a Rock Suitability Classification system (RSC) that can be used to reduce the uncertainties in conditions to which canisters are exposed. A robust disposal system is usually also based on a design that is fairly simple and that works in a transparent and predictable fashion. Some uncertainties will always remain and have to be assessed in terms of their relevance to the final conclusions on safety. The treatment of uncertainty in TURVA assessments is outlined in Section and is further illustrated in Chapters 6, 7 and 8. The combination of deterministic and probabilistic approaches allows the impact of uncertainties on system performance and safety to be determined individually and jointly. This allows uncertainties that could potentially weaken the safety case to be identified, and avoided or reduced by further research and technical development. One of the outputs of TURVA-2012 is the identification of key safety related issues to be addressed in future RTD studies (Section 10.2).

105 60 Figure 2-7. Iterative approach to the management of uncertainties. 2.5 Quality management Goals and principles Posiva applies a quality management system that complies with the ISO 9001:2008 standard for all activities including the production of the safety case reports, and requires the pursuit of the same quality assurance principles from all its contractors and suppliers. The system was first launched in 1997 and has since been subject to continuous maintenance, updating and several internal and external audits. The purpose of Posiva s management system is to ensure, in a documented and traceable way, that Posiva's products whether in the form of abstract knowledge and information, published reports or physical objects fulfil the requirements set for them.

106 61 The general quality objectives, requirements and instructions defined in Posiva s management system will also form the foundation for the quality management of safety case activities carried out in the future. However, special attention is paid to the management of the processes that are applied to produce the safety case and its basis. The purpose of this enhanced process control is to offer full traceability and transparency of the data, assumptions, modelling and calculations. While the approach is based on the ISO 9001:2008-standard, which means management through processes, the principle of a graded approach, as proposed in the safety guides for nuclear facilities, is pursued in safety case production. The graded approach means that the primary emphasis in the quality control and assurance of safety case activities is placed on those parts of the assessment that have a direct bearing on the arguments and conclusions on the long-term safety of disposal, whereas standard quality measures are applied in the supporting work Application to TURVA-2012 safety case production The overall plan, main goals and constraints for the TURVA-2012 safety case production process are presented in the Safety Case Plan (Posiva 2008). The details of how the Safety Case Plan 2008 is being implemented are described in the SAFCA project plan. The work is managed and coordinated by a SAFCA core group and supervised by a steering group. A SAFCA quality co-ordinator (QC) has been designated for the activities related to the quality assurance measures applied to the production of the safety case contents. The QC is responsible for checking that the instructions and guidelines are followed and improvements are made in the process as deemed useful or necessary. The QC is also responsible for coordination of the external expert reviews, maintenance of schedules, review and approval of products, and management of the expert elicitation process. The QC also leads the quality review of models and data used in the Data Handling and Modelling subprocess. Regular auditing of the safety case production process is done as part of Posiva s internal audit programme. The production of the safety case is divided into four main subprocesses: Conceptualisation and Methodology, Data Handling and Modelling, Safety Assessment, and Evaluation of Compliance and Confidence. The Conceptualisation and Methodology subprocess frames the assessment providing the description of the disposal system, FEP analysis and the formulation of scenarios, including system evolution. It guides the definition of the performance targets for the EBS and the bedrock, which form the core of the requirements management system (VAHA). An approach to evaluating the suitability of the rock at various scales has been developed through application of the rock suitability classification (RSC) system. The Data Handling and Modelling subprocess identifies the lines of evolution that could lead to the release of radionuclides and formulates the scenarios that are analysed first to quantify the releases from the repository system to the surface environment and then to quantify the radiological impact of those releases to humans, plants and animals.

107 62 The Safety Assessment subprocess identifies the lines of evolution that could lead to the release of radionuclides and formulates the scenarios that are analysed first to quantify the releases from the repository system to the surface environment and then to quantify the radiological impact of those releases to humans, plants and animals. The Evaluation of Compliance and Confidence subprocess is responsible for the final evaluation of compliance of the assessment results with the regulatory criteria and the overall confidence in the safety case, taking into account the completeness of the scenarios considered, uncertainties within the assessment and complementary considerations regarding the long-term safety of geological disposal. It is essential that the information and data passed between subprocesses is quality assured. Models and Data for the Repository System and Biosphere Data Basis and Biosphere Assessment act as the quality assured interface between the safety case activities and the research and technical activities: they include all the essential EBS and site information and data needed for the performance and safety assessment calculations, while more details can be found in the supporting background reports, such as Site Description and various Production Lines reports. The quality of Site Description is mainly ensured by the application of scientific principles, while the methods of quality control for the design and implementation depend on the nature of the materials and technology in question Model qualification and code verification A range of quality control and assurance measures has been adopted to promote confidence in the models and codes and hence to promote confidence in the analysis of the calculation cases. According to Posiva s Safety Case Plan (Posiva 2008), the quality control and assurance measures comprise: 1. validation of input data for the scenarios and models considered; the limits of applicability of the input data are checked against the assumptions related to the scenarios and models, 2. validation of the models used to analyse the scenarios, 3. verification of assessment codes, 4. validation of the assessment codes for the intended application, 5. documentation of input for the production runs, 6. application of a procedure to ensure codes are correctly applied, 7. documentation of the code versions used, and 8. reporting of non-conformities. Measures 1 and 2 relate to the quality of models and to the selection and checking of data that are implemented in the codes. Actions undertaken to validate and promote confidence in the models and data used in TURVA-2012 are described in Models and Data for the Repository System and for the surface environment in Terrain and Ecosystems Development Modelling, Surface and Near-Surface Hydrological Modelling, Biosphere Radionuclide Transport and Dose Assessment and Dose Assessment for Plants and Animals.

108 63 At a more general level, Features, Events and Processes and Complementary Considerations summarise the understanding of processes relevant to repository performance and safety that can be gained from observations at the site, including its regional geological environment, and from natural and anthropogenic analogues for the repository and its components. Measures 3 to 8 relate to the selection, testing and application of computer codes used for the radionuclide release and transport calculations as well as dose assessment and to the documentation of results. Actions undertaken to verify and promote confidence in the computer codes and their application are described in Assessment of Radionuclide Release Scenarios for the Repository System and Biosphere Assessment. Verification measures, including benchmarking exercises that address specific functions of GoldSim and MARFA, have been carried out during the development of these codes. In addition, benchmarking exercises have been carried out in which results generated by these codes were compared with those generated by REPCOM and FTRANS, which are the codes that were used in previous Posiva safety analyses and have been shown to handle the main features, events and processes of relevance. The benchmarking exercises used test cases that are representative of the types of calculations for which GoldSim and MARFA are used in the TURVA-2012, and so contribute to validation as well as verification. Finally, an external review of MARFA has been carried out and deficiencies identified in the review were addressed before the calculations for TURVA-2012 were undertaken. Based on all these measures, it is concluded that GoldSim and MARFA have been verified and validated to the extent required for use in TURVA Regarding code application, numerical parameters, such as the size of the time steps used by GoldSim and the number of particles calculated by MARFA, are carefully selected to ensure that the model results are sufficiently accurate. A version management system (VMS) has been used to keep track of any changes in input files and thus maintain the reproducibility of calculation results. An assessment database has been set up for the storage, checking and exchange of input data, intermediate results and final results. Finally, an electronic system docgen 12 has been developed to keep track of, and to archive, the results of safety assessment calculations as they are produced. Results of model calculations and their associated input files are downloaded to docgen automatically from the assessment database. In this way, it has been possible to follow the progress of the calculations and carry out quality assurance and plausibility checks in a timely manner Data clearance A wide variety of data have been used for the compilation of the safety case. An important activity for ensuring the quality, transparency and consistency of the data used in the safety case is data clearance. Data clearance is the formal procedure to approve the data to be used as input to the models used in the analyses reported in the safety case, such as the assessment of the performance of the repository system, the analysis of the release scenarios and the analysis of radiological consequences. 12 The docgen system was originally developed for Nagra, the Swiss National Cooperative for the Disposal of Radioactive Waste. The version used in TURVA-2012 has been extended and adapted for Posiva.

109 64 The data produced, e.g. by site investigations or laboratory tests, are usually not directly suitable for the models used in the safety case, and further interpretation and modelling are needed. Sometimes there are no site-specific data available, thus literature data and data from other sources, e.g. data produced by other nuclear waste management organisations, need to be used. The applicability of the data for the specific purpose and conditions analysed in the safety case is assessed as part of the data clearance process, and potential sensitivity cases to be addressed by modelling are suggested. The data may be in the form of single parameter values, a range of parameter values or a probability distribution function. In some cases, different data apply to specific model variants or versions (e.g. applying to a specific hydrogeological model or repository layout). The data clearance procedure consists of the following steps: (i) identification of the data needs, (ii) collection of suitable data, (iii) documentation of the suggested data, their intended use and justification for their selection and (iv) data approval. Separate reports on various categories of data collection have been produced, e.g. regarding climate evolution, solubility and sorption data for the near field and far field, and earthquake frequency. Further review of the data by subject matter experts and safety analysts has been carried out and, in some specific cases a formal expert elicitation has been applied. Purpose-specific databases have been applied to manage the data clearance procedure in a structured way and to ensure the controlled use and traceability of input data used as input to safety related assessment calculations. The expert elicitation process has been applied to a specific case (solubility and sorption data) to identify the main sources of uncertainty and determine whether different views may have to be propagated through the safety assessment. This expert elicitation process has been initiated, recruited, documented and managed by the SAFCA Quality Co-ordinator. The clearance procedure is documented in Models and Data for the Repository System, Biosphere Data Basis and Biosphere Assessment. These reports give an overview of the modelling carried out within the safety case and how the different models link to each other. They also present the key models and data used in the safety case. For each model, the conceptual model, the numerical model and the codes used are described. This description covers the key assumptions and simplification, e.g. omission of certain processes. The uncertainties related to modelling and their impact on the results is presented. Also, possible alternative models are discussed and the basis for selection of the specific model is given. Discussion of the data describes the production, qualification and uncertainties related to the data as well as potential alternative data. In order to assess confidence in the models and data, the applicability of the models and data to the conditions at Olkiluoto and to the safety case purposes as well as the applied quality measures are discussed. Further, the impact of the uncertainties in the models and data on the modelling outcome is assessed and needs for model and data improvements are identified, if necessary. The clearance process has been implemented according to guidelines that address the documentation of data sources and quality aspects. Single items of data and databases are approved through a clearance procedure supervised by the SAFCA Quality Co-

110 65 ordinator. Process owners check and approve the data and while the Quality Coordinator checks and approves the procedure. Data used may also be approved using other Posiva databases like VAHA or POTTI and the respective approval processes. A clearance procedure has been applied to all key data used in the performance assessment (i.e. showing compliance with performance targets and target properties), and in safety assessment (i.e. radionuclide transport analysis and dose calculations) Report and product review and approval process The review and approval of the safety case products (i.e. main portfolio reports) has been done in a fully traceable manner. This has included, first, an internal review by safety case experts and subject-matter experts within Posiva s RTD programme and, second, a review by external experts. A group of external experts covering the essential areas of knowledge and expertise needed in safety case production has been set up. The review comments are managed via review templates, which record the review comments and how each comment has been addressed. Upon completion, this template is checked and approved according to the quality guidelines of Posiva. Quality assurance and quality control measures related to the production and operation of the repository are discussed in detail in Production Line reports (Canister, Buffer, Backfill, Closure and Underground Openings Production Line reports).

111 66

112 67 3 DESCRIPTION OF THE DISPOSAL SYSTEM This chapter presents a summary description of the disposal system in its initial state; that is, descriptions of the host rock and surface environment, and of the spent nuclear fuel and engineered barriers (canister, buffer, backfill and closure). The disposal system (Figure 2-4) is composed of: the spent nuclear fuel, which is the source of hazard to be isolated and contained; the barriers, which are the engineered barriers that are designed, and host rock that is chosen, to isolate and contain the waste, and that are subject to performance requirements (see Section 2.2); the surface environment, which is the environment to be protected. In the TURVA-2012 safety case, the disposal system is described first in its initial state. This is presented in Description of the Disposal System, wherein the initial state 13 is defined as: the state when the direct control over that specific part of the system ceases and only limited information can be made available on the subsequent development of conditions in that part of the system or its near-field. Description of the Disposal System provides a compilation of information presented in more detail in various background reports, the most important being the Canister, Buffer, Backfill, Closure and Underground Openings Production Line reports, in respect of the engineered system, and in Site Description and Biosphere Description, in respect of the natural system. These descriptions provide the basis for considering the performance of the repository system within Performance Assessment (Chapter 6 in this report) and the release of radionuclides and calculation of doses in Assessment of Radionuclide Release Scenarios for the Repository System and Biosphere Assessment (Chapter 8 in this report). 3.1 Host rock The initial state for the host rock is defined to be the baseline conditions prior to starting the construction of the ONKALO. The Olkiluoto site, as seen today, is the consequence of events and processes that have taken place over billions of years, from those reflected in the geological properties of the rocks forming the geosphere, to the shorter-term changes related to climate-driven processes that mainly cause changes in groundwater flow and groundwater composition and the geomechanical response to crustal movements related to glacial loading and unloading. The present state and the past history of the geology, rock mechanics and thermal properties, hydrogeology and hydrogeochemistry of Olkiluoto are discussed in detail in Site Description. The repository is to be excavated on Olkiluoto Island in south-western Finland (Figure 2-2). The crystalline bedrock of Finland is a part of the Precambrian Fennoscandian 13 Note that the definition is somewhat different for the surface environmnet and the host rock (see text).

113 68 Shield, which, in south-western Finland belongs to the Svecofennian domain, which developed between 1930 Ma and 1800 Ma ago. The rocks of Olkiluoto consist of two major classes: high-grade metamorphic rocks including gneisses with varying degree of migmatisation, and igneous rocks including pegmatitic granites and diabase dykes (Figure 3-1). The bedrock has been affected by five stages of ductile deformation resulting in lithological layering, foliation, and strong migmatisation and folding. Extensive hydrothermal alteration has also affected the properties of fractures and certain rock volumes, the main alteration minerals being illite, kaolinite, sulphides and calcite. As a result, the rock properties at Olkiluoto are heterogeneous, which is reflected also in the variation of the thermal and rock mechanics properties and seen for example in the anisotropic thermal properties due to foliation and gneissic banding. The fault zones at Olkiluoto are mainly SE-dipping thrust faults formed during the latest stages of the Svecofennian orogeny, approximately 1800 Ma ago, and were reactivated in several deformation phases (see Figure 3-1 and Figure 3-2). In addition, NE-SW striking strike-slip faults are also common. The occurrence of fracturing varies between different rock domains, but the following three fracture sets are typical for the site: (i) east-west striking fractures with generally subvertical dips to both the north and south, (ii) north-south striking fractures with generally subvertical dips to both the east and the west and (iii) moderately-dipping to gently-dipping fractures with strikes that are generally sub-parallel to the aggregate foliation directions in a particular fracture domain. In Fennoscandia, the orientation of the major principal stress is attributed to an E-W compression from the mid-atlantic ridge push and a N-S compression from the Alpine margin, resulting in a roughly NW-SE orientation of major principal stress (Heidbach et al. 2008). This is also supported by the regional in situ data from Olkiluoto and other Finnish sites studied during the site selection programme. Changes in isostatic load due to glaciations and related isostatic adjustment and the existence of brittle deformation zones change the stress regime at the site. Currently, a thrust faulting stress regime is present, i.e. the horizontal stresses are larger than the vertical stress, H > h > v and the principal stresses are approximately oriented horizontally and vertically, respectively. The orientation of H at the site is found to vary slightly with depth and at the repository depth is in the range NW-SE to E-W. The vertical stress is generally close to that expected due to the weight of the overlying rock. Located away from active plate margins, Fennoscandia, and Olkiluoto in particular is known as a seismically quiescent region. Increased seismicity in Fennoscandia is possible in connection to the most recent glaciation and post-glacial faults have been discovered in northern Fennoscandia (e.g. Kuivamäki et al. 1998). There are no direct signs of post-glacial faulting in the vicinity of Olkiluoto (e.g. Lindberg 2007) although disturbances of the sea-bottom sediments have been suggested to be related to postglacial faulting (Hutri et al. 2007). According to the data from historical earthquakes, the Olkiluoto area is located within a zone of lower seismicity, the Southern Finland Quiet Zone (SFQZ), between two seismically active belts, Åland Paldis Pskov (Å-P-P)

114 69 Figure 3-1. A geological map of Olkiluoto Island showing the lithology and the brittle fault zones (BFZ) defined as layout determining features, i.e. the ones that restrict the repository layout. Figure 3-2. Three-dimensional representation of the main hydrogeological zones at Olkiluoto (HZ in blue) and their correlation with the fault zones (BFZ, in red) (outline of the island is shown in the figure, oblique view towards the northeast).

115 70 and Bothnian Bay Ladoga (B-L) (see Figure 2-2 in Saari 2008). These seismically active zones seem to be essential elements when the driving mechanisms of the seismicity of southern Finland are regarded. The zones are distinguished from their surroundings particularly by the occurrence of relatively large (M 3.5) earthquakes. In the crystalline bedrock at Olkiluoto, groundwater flow takes place in hydraulically active deformation zones (hydrogeological zones) and fractures. The larger-scale hydrogeological zones, which are related to brittle fault zones, carry most of the volumetric water flow in the deep bedrock. There is a general decrease in the frequency of transmissive fractures, and of transmissivity of both fractures and the hydrogeological zones with depth. Under natural conditions, groundwater flow at Olkiluoto occurs mainly as a response to freshwater infiltration dependent on the topography, although salinity (density) variation driven flow also takes place to a lesser extent. The porewater within the rock matrix is stagnant but exchanges solutes by diffusion with the flowing groundwater in the fractures. The distribution of the groundwater types is the result of progressive mixing of groundwaters and the slow interaction between the groundwater, porewater and the minerals of the rocks (see Figure 3-3 and Site Description). The groundwater composition is also affected by microbial activity. Water-rock interactions, such as carbon and sulphur cycling and silicate reactions, buffer the ph and redox conditions and stabilise the groundwater chemistry. Weathering processes during infiltration play a major role in determining the shallow groundwater composition. Pyrite and other iron sulphides are common in waterconducting fractures throughout the investigated depth zone indicating a strong lithological buffer against oxic waters over geological time scales. Groundwaters, in the range down to 300 m depth show indications of having been affected by infiltrating waters of glacial, marine and meteoric origin during the alternating periods of glaciations and interglacials during the Quaternary. On the other hand, these indications are absent in fracture groundwaters below 300 m, implying that these groundwaters are older. The current fracture groundwater is characterised by a significant, variation of salinity with depth (see Figure 3-3). Fresh waters (<1 g/l) rich in dissolved carbonate are found at shallow depths, in the uppermost tens of metres. Brackish groundwater, with salinity up to 10 g/l dominates at depths between 30 m and about 400 m. Sulphate-rich waters are common in the depth layer m, whereas brackish chloride water, poor in sulphate dominates at depths of m. Saline groundwaters (salinity >10 g/l) dominate at still greater depths. The matrix porewaters seem to be in equilibrium with the fracture groundwaters in the upper part of the bedrock (0 150 m), suggesting similar origin and strong interaction between groundwater in fractures and matrix at these depths. At deeper levels ( m), the matrix porewater is less saline and increasingly enriched in δ 18 O; this has been interpreted to represent fresh water conditions during a warm climate, probably during the preglacial Tertiary period, anion exclusion being another possible explanation (Posiva 2009b).

116 71 Figure 3-3. Illustrative hydrogeochemical site model of baseline groundwater conditions with the main water-rock interactions at Olkiluoto. Changes in colour indicate alterations in water type. The hydrogeologically most significant zones are represented. Blue arrows represent flow directions. Rounded rectangles contain the main sources and sinks affecting ph and redox conditions. Enhanced chemical reactions dominate the infiltration zone at shallow depths, and at the interface between Na-Cl-SO4 and Na-Cl groundwater types. The illustration depicts hydrogeochemical conditions in the water-conductive fracture system, not in the diffusion-dominated rock matrix (Site Description). 3.2 Surface environment The surface environment (including the overburden and the surface hydrogeology and hydrogeochemistry) is described in detail in Biosphere Description. The initial state of the surface environment corresponds to the description of the present-day conditions. Topography in the Olkiluoto area, and in general in south-western Finland, is flat and soil erosion rates are very low. Glacial erosion features such as glacially smoothed

117 72 bedrock outcrops and roches moutonnées 14 are common. As a result of the last glaciation, the bedrock depressions are filled with a thicker layer of overburden, mainly sandy till and fine-grained till (e.g. Lahdenperä 2009 and Biosphere Description). The sea around Olkiluoto Island is shallow, except for a few areas where water reaches a depth of about 15 m. The seabed deposits in the surroundings of Olkiluoto are heterogeneous and sediment thickness is variable. About % of the offshore is covered by till. Exposed bedrock and sedimentary rock form about % of the area of the seabed and various kinds of soft sediments cover about % in the deeper open sea area and in sheltered near-shore basins (Rantataro & Kaskela 2009, Ch. 5). Littorina clays have been deposited in a marine environment, when environmental conditions in the sea were favourable for a moderately abundant fauna. Because of the amount of sedimented and intact organic material, these clays are gyttja clay in which there has often been gas formation, inside the sediment, as a consequence of the breakdown of organic material (Rantataro & Kaskela 2009, p. 13). This is of particular interest because of the continued land uplift of the Olkiluoto area which will expose areas of current seabed sediments in the next few thousand years the time window for which the radionuclide releases must be quantified in terms of annual doses to humans and absorbed dose rates to other biota. In addition, the effects of land uplift are accentuated by paludification e.g. reedbed growth in the coastal areas, especially in shallow bays (Haapanen & Lahdenperä 2011). The ecosystem succession during uplift, and the redistribution of sediments and groundwater flow, will influence the areas of potential deep groundwater recharge and discharge from the repository (Haapanen et al. 2007, 2009). The net changes in sea level are a result of both crustal uplift and changes in sea level due to changes in climate globally. The effect of uplift could be, at least to some extent, enhanced or reversed by changes in sea level. This is taken into account in the climate scenarios in Biosphere Assessment. At present, freshwater (limnic) ecosystems are few in the Olkiluoto area and there are no natural lakes on the island. The nearby lake basins were isolated during the various stages of development of the Baltic Sea as a result of isostatic uplift and tilting of the land. The closest rivers are the Eurajoki and Lapijoki, which discharge to the sea to the north and east of Olkiluoto, increasing the concentration of nutrients and solids, especially at the river mouths (Haapanen et al. 2009). The use of lakes and mires currently absent from the island in the surrounding region as analogues for future biosphere conditions at Olkiluoto is discussed by Haapanen et al. (2010). Figure 3-4 shows land uplift and an example of surface environment development through to 10,000 years from the present. 14 Outcrop of hard rock that has been shaped by the action of a glacial movement and erosion. The roche typically has a low, smooth, rounded end pointing upstream, with reference to the direction of ice movement, and a higher, rougher, ice-plucked, downstream end. The surfaces may be marked by glacial striations.

118 73 Figure 3-4. Land uplift and an example of the biosphere development through to 10,000 a after the present. Map data: Topographic database by the National Land Survey of Finland (permission 41/MYY/11) and Posiva Oy. Map layout by Jani Helin/Posiva Oy. Note: dates are given as AD, i.e is 10,000 years after the reference date of 2020 AD.

119 Underground openings and repository layout Underground openings of the disposal system include all spaces excavated underground, including access tunnel, shafts, technical and demonstration rooms, central tunnels, deposition tunnels and deposition holes. Drillholes within the area are also included. Underground openings are constructed, utilised and backfilled in a stepwise manner during the course of the disposal operation. The initial state of the underground openings is the state of rock at the time of emplacement of the material (e.g. buffer, backfill) intended for a specific space. Rock Suitability Classification criteria (RSC criteria) (McEwen et al. 2013) are applied during construction (Underground Openings Production Line) to achieve conditions that meet the target properties set for the host rock in Design Basis. The drill and blast method is used in the construction of tunnels and raise boring in shafts. To limit the extent and connectivity of the EDZ, the deposition holes will be bored. To control groundwater inflow in the tunnels, grouting has been and will be used when necessary during construction 15 (for estimates of the grout quantities, see Karvonen 2011), but grouting will not be allowed in the deposition holes. At the repository level, central tunnels lead to deposition tunnels, in which the deposition holes are located. Defining locations for these spaces follows the procedures being set by the RSC system (McEwen et al. 2013, see also Section 2.2.3). The final placing of the deposition tunnels and subsequently deposition holes is determined by exploratory means before construction. The layout of the underground openings is constrained by the layout-determining features (LDFs), which are large lineaments, significant brittle fault zones (BFZ) or hydrogeological zones (HZ) (McEwen et al. 2013). The layout used in the TURVA safety case is presented in Figure 3-5 (Saanio et al. 2013). There is flexibility to adapt the layout according to the additional and more detailed geological information to be gained during construction. The basic statistics for the reference layout are summarised in Table 3-1; deposition tunnel and hole dimensions depend on the fuel type to be disposed. The layout provides for a total of 5400 positions for deposition holes compared to 4500 actually required. This allows for rejection of some sections of deposition tunnels and some deposition holes that do not meet the RSC criteria. Several layout adaptations of the repository have been produced for a repository to host either 5500 (Kirkkomäki 2009), 5440 or 9000 tu (Saanio et al. 2013). The final layout will be adjusted in future taking into account the findings of the continued site characterisation and possible other constraints (e.g. land use restrictions). The current reference layout in the construction licence application is presented in Figure 3-6. Provided that the deposition tunnels and holes are located so that major fault zones and hydrogeological zones are avoided, the findings of the safety case are not sensitive to details of the layout. 15 Low ph grouts are recommended to be used in ONKALO and are also required to be used as primary grouting material below the depth of 300 m. Low ph grouts are used to avoid ph increase in groundwater. In addition, at repository level non-cementitious grouts will be favoured where possible.

120 75 Foreign materials will be introduced during the construction and operation and, although most will be removed and significant spills contained, inevitably some foreign materials will be left in the repository. The amounts of foreign materials (e.g. cementitious materials) introduced are being followed and controlled (including the imposition of limitations) at the construction site and the total amounts are estimated regularly (see Karvonen 2011 for the latest estimates). Table 3-1. Layout statistics following the Description of the Disposal System report. Common layout parameters Total amount of the fuel to be disposed, tu 9000 Total number of canisters to be disposed 4500 Total number of canister locations in the layout 5400 Repository depth (metres below ground) 400 to 450 parallel to the main horizontal stress, which Orientation of the deposition tunnels is at repository level in the range NW-SE and E-W. Maximum length of deposition tunnel (metres) 350 Deposition hole diameter (mm) 1750 Fuel type specific parameters for: OL1 2 OL3 4 LO1 2 Number of canisters Nominal cross-section of deposition tunnels (m 2 ) (OL3) 12.7 Deposition hole height (metres) Distance between deposition holes (metres)

121 76 Figure 3-5. Layout adaptation for a repository hosting 9000 tu of spent nuclear fuel used in the TURVA-2012 safety case, dark grey areas are not suitable for deposition tunnels according to the RSC as they are intersected by LDFs and their respect volumes. Red ovals denote respect distances to drillholes (Saanio et al. 2013). Figure 3-6. The current reference layout (green). Grey areas are not suitable for deposition tunnels based on Rock Suitability Classification (RSC). Red ovals denote respect distances to drillholes. Red line surrounding the repository shows the area reserved for the repository in urban planning.

122 Spent nuclear fuel The spent nuclear fuel produced by the currently operating reactors, the OL3 unit under construction and the planned OL4 unit are each different depending on the reactor type. The OL1 and OL2 reactors at Olkiluoto are boiling water reactors (BWR), Loviisa LO1 and LO2 are VVER-440 type reactors and OL3, currently under construction, will be a pressurised water reactor (PWR, trade name EPR). The design of the fuel assemblies varies depending on the reactor type. The OL4 reactor type has not been decided yet and in the TURVA-2012 safety case it is assumed to correspond to OL3. The initial state of the spent fuel is described in Chapter 5 of Description of the Disposal System. The spent nuclear fuel comes from storage at the nuclear power plants as assemblies that contain the spent fuel pellets within alloy tubes and other metal parts comprising the assembly. Figure 3-7 provides illustrations of different fuel assemblies. For each fuel type, the canister design is adapted to accommodate the spent fuel to be disposed (see Section 3.5 and Figure 3-8), in particular in terms of overall length and insert. Figure 3-7. Representative illustrations (from left) LO1 2, OL1 2 and OL3 type fuel assemblies. LO1 2 and OL1 2 fuel elements are partly cut open to show the internal structures. The pictures are not to scale.

123 78 The spent fuel pellets from all the Finnish reactor types are, from a chemical point of view, made of the same sintered material, UO 2. However, the UO 2 pellet geometry, U-235 content and burnable poison (absorber) proportion, and cladding material, as well as other components of the assembly, are different depending on the reactor type. The average U-235 enrichment of a fuel assembly may vary roughly between 3 4 %, but within a single fuel rod/pellet could be nearly 5 %. In the future, the assembly average enrichment rate could be over 4 %, enabling higher burn-ups to be achieved. Fuel cladding is made of various types of zirconium alloys because the zirconium crosssection for thermal neutrons is very small and because zirconium alloys typically have great mechanical strength and good corrosion resistance. The other structural elements of the fuel assemblies (i.e. upper and lower tie-plates, end plug, spacer grid, and channel and nose piece) are fabricated from stainless steel, zirconium alloys or nickel-based alloy). For more details, see Description of the Disposal System. The most important properties that are considered are the material properties of the assemblies, degree of burn-up, heat output levels and radionuclide inventory, which are defined for the initial state (see Description of the Disposal System, Chapter 5). Some basic characteristics of the fuel types are listed in Table Canister The initial state of a single canister is the state when the canister filled with spent fuel has been emplaced in a deposition hole, the surrounding buffer is present (Section 3.6) and the deposition tunnel backfill has been emplaced on top of the deposition hole (Section 3.7). The detailed design and initial state of the canister are described in Description of the Disposal System and in Canister Production Line. Due to the several spent fuel types (see Section 3.4 and Chapter 5 in Description of the Disposal System for fuel details) there are geometrical differences between the canister types (Figure 3-8a). The canister is composed of a cast iron insert and a copper overpack (Figure 3-8b). All canister types have an external diameter of 1.05 m; heights vary between 3.5 and 5.2 m. Table 3-2. Representative fuel characteristics for OL1 2, LO1 2 and OL3 fuels (per assembly) (Canister Production Line report). Fuel type OL1 2 LO1 2 OL3 Mass of uranium (kg) Anticipated maximum average burn-up of 50 a fuel assembly (MWd/kgU) Estimated average burn-up of all the fuel (MWd/kgU) Typical enrichment U-235 (%) Minimum cooling time of a single assembly (years) Minimum average cooling time with average burn-up (years) Allowable average decay heat at disposal (full canisters) (W/tU)

124 79 The material for the copper components is phosphorus-alloyed oxygen-free copper with the following requirements: O <5 ppm, P ppm, H <0.6 ppm, S <8 ppm. Creep testing of Cu-OF (oxygen-free copper) doped with 30 to 120 ppm phosphorus has shown higher creep strength and much better creep ductility than copper without phosphorus according to Andersson-Östling & Sandström (2009, Section 12). The cast iron material composition of the insert is specified only with respect to an upper limit on the content of copper to avoid the risk of radiation embrittlement. The content of copper shall therefore not exceed 0.05 %. During the development of the casting process for the nodular cast iron inserts, the standard requirements in EN 1563 grade EN-GJS U have been used regarding mechanical properties (Raiko et al. 2010). Dose rates outside the canister, canister temperatures, presence and composition of water and gas in the canister, the type and probability of initial penetrating defects as well as the residual welding stresses at the initial state are presented in Chapter 6 of the Description of the Disposal System. The design aim for the canisters is that all canisters are intact at emplacement and will remain intact for hundreds thousands of years. It cannot be ruled out, however, that one or few canisters with an initial penetrating defect may be emplaced (as discussed in the Canister Production Line report). The cautious assumption that one or a few canisters may have initial penetrating defects in the weld is based on expert judgement concerning the canister welding method (electron beam welding EBW) and nondestructive testing (NDT) capabilities. With more data becoming available in the future, it is likely that it will be possible to demonstrate that the probability of emplacing more than one canister with an initial undetected penetrating defect is less than one per cent. At the moment, therefore, the number of defective canisters is assumed to be one canister out of 4500 in the formulation of release scenarios; e.g. the reference case realisation of the base scenario (see Chapter 7). Figure 3-8. a) Canister geometries from left to right for spent fuels from LO1 2, OL1 2 and OL3 (OL4). b) Canister overpack is made out of copper and the insert is cast iron, spent fuel rods are placed in the channels in the insert.

125 Buffer The reference buffer material is high grade Na-bentonite from Wyoming (MX-80). Other similar materials can be considered in the future as long as they meet the design requirements. The typical mineralogical composition for the buffer bentonite is given in Table 7-3 in Description of the Disposal System. The compacted rings and discs of buffer will be emplaced in the deposition hole that is bored and accepted according to the RSC criteria. The canister will be emplaced within the buffer as illustrated in Figure 3-9. The bottom of the deposition hole will be smoothed to ensure a tight contact between the host rock and the lowermost buffer disk. The inner gap between the buffer and the canister is unfilled and will present a small air gap. The outer gap between the buffer blocks and rock will be filled with pellets of buffer material. The buffer is in contact with the deposition tunnel floor backfill material. Detailed dimensions of the buffer depend on the canister type to be emplaced (see Figure 3-9) (also see Table 7-1 in Description of the Disposal System for buffer dimensions). The initial water content in the buffer will be 17 %. The porosity of the rings surrounding the canister is to be 36.0 % and the porosity of the discs on top and below the canister is to be 38.2 %. Detailed buffer properties at the initial state including the amounts of air, oxygen and water present, as well as impurities, are presented in Description of the Disposal System (in Chapter 7, Table 7-2). Figure 3-9. Illustration of the emplacement of the buffer and canister in the deposition hole and placing the backfill on top (background figure).the buffer designs (three designs depending on the spent fuel) are presented as schematic figures on left (from left to right LO1 2, OL1 2 and OL3).

126 Backfill and plug At the initial state of the deposition tunnel backfill both the clay backfill and the deposition tunnel plug have been installed as illustrated in a schematic Figure The layout of the deposition tunnel backfill varies due to the different deposition tunnel sizes within the repository, which are described in more detail in the Backfill Production Line report and in Description of the Disposal System, Chapter Deposition tunnel backfill The backfill components (blocks, foundation layer and pellets) will be emplaced in the deposition tunnels as illustrated in Figure The main backfill material in the reference design is Friedland clay (blocks). Bentonite clay materials are used for pellet fill surrounding the blocks (Cebogel pellets) and for the foundation layer to be emplaced on the tunnel floor (Minelco granules). The foundation layer smoothes out the unevenness of the drill and blast excavated floor in the tunnels allowing the emplacement of the blocks. Figure Schematic figures showing (upper) the main backfill components in the deposition tunnel and (lower) a cross-section of the OL1 3 tunnel with the schematic presentation of theoretical excavation extent (min and max with dashed lines) and a possible realisation of the backfilling components installed in the tunnel (Backfill Production Line report).

127 82 Backfill has been designed taking into account the tolerances in excavation (Figure 3-10 lower figure). Prior to saturation, the average degree of saturation (ratio of volume of water to volume of voids) is 55 % for the backfill. The rest of the void volume is filled with air (45 %). Detailed descriptions of the backfill materials, their composition and other properties at initial state are given in Chapter 8 of Description of the Disposal System Deposition tunnel plug The current deposition tunnel plug design is based on SKB s plug design (see SKB 2010a) (see Figure 3-10 upper figure). It consists of a concrete dome; bentonite sealing layer and a sand filter (see Description of the Disposal System for more details). The combination structure ensures that the plug has sufficient hydraulic isolation capacity as well as structural strength. However, the plug design is still at a conceptual level and tests are needed to verify its hydraulic isolation capacity. In addition, the formulation of the concrete mix is under development and may change in the future. The concrete components (concrete plug and beams) in the deposition tunnel end plug are made of low ph concrete (a concrete with a ph of the leachate < 10, with a short period of initially higher ph of about 11), such as presented in the Backfill Production Line report. For the formulation currently adopted, the water to cement ratio is (kg/kg), water to binder ratio is (kg/kg) and water to dry material ratio is 0.29 (kg/kg). The installation of deposition tunnel plug allows the use of the central tunnels during the operational phase. Thus, the hydraulic requirements set for the plug are only for the operational period (see Table 2-2, L3-BAC-9 and L3-BAC-18). The previous design by Haaramo & Lehtonen (2009) has been used as reference geometry in the TURVA-2012 safety case (see Description of the Disposal System for more details). 3.8 Closure Closure will complete the isolation of the waste, restore and maintain favourable natural conditions in the bedrock, and prevent the formation of preferential flow paths and transport routes between ground surface and deposition tunnels and holes. Closure of the disposal facility covers all backfilling and plugs outside the deposition tunnels, including sealing of drillholes. The current reference design for the closure of the disposal facility is presented in Figure The detailed description of the reference design, the backfill materials and methods, as well as the principles of the hydraulic, mechanical and intrusion-obstructing plugs are given in Closure Production Line report.

128 83 Figure Current reference design for closure showing the access tunnel (1) and shafts (2), technical rooms at the repository level (3) and the central tunnels (4) leading from the technical rooms to deposition tunnels, location of the L/ILW repository is also shown (5) (Closure Production Line report). The reference design for closure will deploy a flexible tool-box of techniques to accomplish closure to the required standards and performance requirements. The available techniques provide alternative solutions throughout the closure process for the emplacement of backfill and plugs in a manner that meets the requirements set. Natural materials are utilised in backfill (such as clays, aggregates and mixtures of these). In plugs, at least below structure HZ20 (a major hydrogeological zone lying above the repository which forms a hydraulic discontinuity see Figure 3-2), low ph concrete will be used. See also Description of the Disposal System (Chapter 9) for the initial state details.

129 84

130 85 4 FEATURES, EVENTS AND PROCESSES This chapter describes the identification and screening of features, events and processes (FEPs) and development of a database of FEPs relevant to the performance assessment and analysis of potential radionuclide releases and radiological impacts. It also outlines the onward use of the FEP descriptions in performance assessment and radiological analyses and modelling, and outlines the future lines of evolution of the disposal system and its environment. 4.1 Identification and screening of FEPs Identification of potentially relevant FEPs Identifying and describing the features, events and processes (FEPs) that are relevant to understanding the evolution of the disposal system, or to its performance and safety, is an essential step towards ensuring comprehensiveness of the assessments and safety case. For the TURVA-2012 safety case, the identification and description of FEPs and couplings between these has been done based on scientific understanding without considering the capabilities of the models that may be used to represent the processes or disposal system components. This is to promote the elicitation of a complete set of relevant FEPs, including FEPs that may not be included in the set of models that are initially available. Following identification of a comprehensive list of FEPs a systematic screening process is applied to rule out those FEPs that cannot be relevant in the context of the repository system as proposed at the Olkiluoto site. The identification and screening of FEPs was carried out by a team of scientific subject and assessment experts. The process was conducted in a structured manner relying on the collaborative (individual and joint) judgements of the experts. This is the fourth iteration of Posiva s Process Report, following from Vieno & Nordman (1997) in support of the TILA-99 assessment, Rasilainen (2004) and Miller & Marcos (2007) in support of the Interim Safety Case 2009 (Posiva 2010). The process took advantage of and drew on the experience from previous Posiva studies, as well as from the development of FEP lists in support of the assessment of the KBS-3V disposal method in Sweden, notably Miller et al. (2002). This is the first iteration also including FEPs for the surface environment. These have in previous assessments mainly been addressed in the biosphere description reports (Haapanen et al. 2007, 2009). An initial long list of FEPs for identification and screening for the TURVA-2012 safety case was derived from the previous Posiva Process Report (Miller & Marcos 2007), the previous Bisophere Description Report (Haapanen et al. 2009), the NEA International FEP list (NEA 1999) and its supporting project database (NEA 2006), together with other relevant safety cases, notably SKB s SR-Can assessment (SKB 2006) SKB s SR-Site assessment became available only after the initial FEP list for Features, Events and Processes was compiled.

131 Screening for relevance to TURVA-2012 The FEPs in this long list were then examined to determine their relevance and potential significance against the following criteria: relevance to the KBS-3V type repository design for spent nuclear fuel disposal; relevance to the present-day Olkiluoto site characteristics and likely future site characteristics evolving in response to climatic changes and other external factors; relevance to the national regulatory requirements and guidelines; previous experience in FEP screening and safety case development by Posiva (see above); knowledge and information gaps identified during the course of Posiva s and SKB s ongoing RTD programmes; the outcomes from previous safety cases for and performance assessments of the KBS-3V type repository design; expert knowledge and awareness of other developing national and international RTD and safety case programmes, and feedback from the regulatory agency (STUK) on previous safety case reports and Posiva s RTD programme. Examples of FEPs that were excluded during the screening process included: Spent fuel degradation due to high-ph waters (ph > 10) was excluded because expert judgement indicates that it is unlikely that significant quantities of high-ph water generated by the chemical degradation of cementitious materials can migrate through the buffer and come into contact with the spent fuel. Reaction between bentonite and backfill and high-ph waters is addressed, however, because of the physical proximity of parts of the buffer [5.2.6] and backfill [6.2.5] to the cementitious plugs at the ends of the deposition tunnels. Deliberate human intrusion was excluded because it is assumed that, if an intrusion is deliberate, appropriate measures would be taken to protect people and the environment. Most of the FEPs that were screened out in the process were also screened out in previous FEP analyses for TILA-96 (Vieno & Nordman 1997) and TILA-99 (Vieno & Nordman 1999). The process of FEP documentation therefore focused on retained FEPs, thus Features, Events and Processes contains descriptions only of those FEPs that passed screening and are considered potentially significant for the long-term safety of the disposal facility.

132 Organisation of the FEPs The FEPs in Features, Events and Processes are organised according to the main components of the disposal system: Spent nuclear fuel; Canister; Buffer; Backfill; Auxiliary components 17 ; Geosphere; Surface environment; the external FEPs are also discussed. For each component, FEPs affecting the evolution the disposal system (termed evolution-related FEPs) and FEPs relevant to radionuclide transport (migrationrelated FEPs) are also distinguished. This leads to the list of retained FEPs as shown in Table Table 4-1. List of retained FEPs organised according to disposal system component as well as the external FEPs. 3. Spent nuclear fuel Radioactive decay (and in-growth) Release of the labile fraction of the inventory Heat generation Production of helium gas Heat transfer Criticality Structural alteration of the fuel pellets Aqueous solubility and speciation Radiolysis of residual water (in an intact Precipitation and co-precipitation canister) Radiolysis of the canister water Sorption Corrosion of cladding tubes and metallic Diffusion in fuel pellets parts of the fuel assembly Alteration and dissolution of the fuel matrix 4. Canister Radiation attenuation Aqueous solubility and speciation Heat transfer Precipitation and co-precipitation Deformation Sorption Thermal expansion of the canister Diffusion Corrosion of the copper overpack Advection Corrosion of the cast iron insert Colloid transport Stress corrosion cracking Gas transport 5. Buffer Heat transfer Freezing and thawing Water uptake and swelling Aqueous solubility and speciation Piping and erosion Precipitation and co-precipitation Chemical erosion Sorption Radiolysis of porewater Diffusion Montmorillonite transformation Advection Alteration of accessory minerals Colloid transport Microbial activity Gas transport 17 Auxiliary components refers to backfilling of central tunnels, service areas, access tunnel and shafts, and seals and plugs that are installed both at the mouths of the deposition tunnels and as part of closure. In Features, Events and Processes, all backfill issues are discussed in the chapter Tunnel backfill, so the chapter Auxiliary components focuses on the seals and plugs (both the deposition tunnel plugs and the closure plugs). 18 The numbering of the FEPs is based on the chapters and sections in Features, Events and Process, where the discussion of FEPs begins in Chapter 3 (Spent nuclear fuel).

133 88 6. Backfill Heat transfer Aqueous solubility and speciation Water uptake and swelling Precipitation and co-precipitation Piping and erosion Sorption Chemical erosion Diffusion Montmorillonite transformation Advection Alteration of accessory minerals Colloid transport Microbial activity Gas transport Freezing and thawing 7. Auxiliary components Chemical degradation Freezing and thawing Physical degradation Transport through auxiliary components 8. Geosphere Heat transfer Microbial activity Stress redistribution Aqueous solubility and speciation Reactivation-displacements along Precipitation and co-precipitation existing fractures Spalling Sorption Creep Diffusion and matrix diffusion Erosion and sedimentation in fractures Groundwater flow and advective transport Rock-water interaction Colloid transport Methane hydrate formation Gas transport Salt exclusion 9. Surface environment Erosion Gas generation Degradation Ingestion of food Podzolisation Inhalation of air Agriculture and aquaculture Respiration Forest and peatland management External radiation from the ground Infiltration Exposure from radiation sources Groundwater discharge and recharge Topography Runoff Well Drainage Construction of a well Capillary rise Food source potential Uptake Dietary profile Evapotranspiration Demographics Translocation Exposed population Litterfall Terrestrialisation Bioturbation Advection Migration of fauna Dispersion Senescence Water exchange Atmospheric deposition Sedimentation and resuspension Atmospheric resuspension Ingestion of drinking water Diffusion Flooding Sorption Water source potential

134 External FEPs Climate evolution Land uplift and depression Glaciation Inadvertent human intrusion Permafrost formation 4.2 Development of the FEP descriptions FEP descriptions The aim of the FEP descriptions (presented in Features, Events and Processes) is to provide a concise summary of the current understanding of each FEP and its associated uncertainties, and to set out when and where it may occur in the disposal system, and how it is coupled to other FEPs. Interactions between disposal system components are also indicated. The descriptions are not intended to define how the FEP is treated in the TURVA-2012 safety case, which is part of the assessment and modelling judgement described in Performance Assessment, Formulation of Radionuclide Release Scenarios, Models and Data for the Repository System, and for the surface environment in Terrain and Ecosystems Development Modelling, Surface and Near-Surface Hydrological Modelling, Biosphere Radionuclide Transport and Dose Assessment and Dose Assessment for Plants and Animals. A description of each of the retained FEPs (Table 4-1) was compiled by a subject matter expert familiar with Posiva s RTD and assessment programme in the relevant topic area. Each description includes the current fundamental scientific understanding and expected relevance in the context of the Posiva disposal system at the Olkiluoto site, e.g. in terms of temporal and spatial occurrence, plus a note of any fundamental uncertainties in scientific understanding. Extensive references are also included. All FEPs are described using a common template with standard fields as defined in Table 4-2.

135 90 Table 4-2. Common template for FEP descriptions. Name and number: a unique name and number are given to each FEP; these are the same as those used in Posiva s developing FEP database. Type: feature, event or process. Class: whether the FEP is system evolution-related or migration-related, or both. (There is not always a clear-cut distinction, but the distinction is useful to indicate whether the FEP is to be considered in performance assessment, radiological assessment or both.) General description: a concise description of the FEP and its consequences for system evolution and safety, covering: current fundamental scientific understanding, any relevant properties, conditions and constraints that affect its operation, the likely temporal and spatial variability of its operation, the conditions under which it may be expected to occur at Olkiluoto, and when in the evolution of the disposal system it is expected to occur. Reference is made to evidence for the FEP from RTD studies in the field or laboratory. Where possible, quantities are given to illustrate the magnitude of the FEP (e.g. process rates), but these are not intended to define the actual parameter values or ranges used in the safety case; these are defined in Models and Data for the Repository System, and for the surface environment, the data used in the biosphere assessment are summarised in Biosphere Data Basis, and the models are discussed in Terrain and Ecosystems Development Modelling, Surface and Near-Surface Hydrological Modelling, Biosphere Radionuclide Transport and Dose Assessment and Dose Assessment for Plants and Animals. Uncertainties in the understanding of the FEP: a short discussion about any uncertainties in conceptual understanding and gaps in knowledge, but generally not about uncertainties in data or modelling approaches; these are defined in Models and Data for the Repository System and Biosphere Data Basis. Couplings to other FEPs: this section lists which FEPs affect or are affected by the FEP being described. References: key references to the main scientific literature and relevant Posiva reports, particularly any recent summary reports on the FEP. This is not intended to be a comprehensive bibliography, but is a signpost to the relevant literature Coupling between FEPs and aggregation/disaggregation Interaction matrices were developed for the FEPs of each main component of the disposal system. These were used to check and identify which FEPs are coupled with each other, and also to highlight which FEPs are most important to controlling radionuclide transport between components of the disposal system. When identifying the potentially significant FEPs, expert judgement was applied when deciding how to address coupled processes. For example, water uptake by the buffer and swelling of the buffer are two separate but connected processes. In Features, Events and Processes and in the retained FEP list (Table 4-1), these two processes are addressed in a single FEP description [5.2.2 Water uptake and swelling] because they are so closely coupled that they can be addressed as a single process. In other cases, closely connected processes have been described separately because they may need to be addressed differently in the development of scenarios and in the performance assessment models. For example, radioactive decay and radiogenic heat generation are very closely related but they each have their own FEP descriptions [3.2.1 and 3.2.2]. In some cases, FEPs have been aggregated or disaggregated differently compared with the previous version of the Process Report (Miller & Marcos 2007). This reflects developing understanding of the roles and importance of different FEPs. The individual

136 91 FEP descriptions include references to features, events and processes in the previous version of the report, for traceability. The process of developing the FEP database and the complete set of descriptions, references and other information is presented in Features, Events and Processes. 4.3 Onward use of the FEP descriptions The initial state of the repository can be described by a set of features that depend on the amount and characteristics of the spent nuclear fuel to be disposed of in the repository, repository design and characteristics of the site. These features are discussed in Description of the Disposal System. As noted above, FEPs are categorised as evolution-related FEPs, which mostly affect the physical state of the disposal system, or migration-related FEPs which mostly affect the release, transport and accumulation of radionuclides (and other chemical species). The synthesis of evolution-related FEPs provides a description of the ways in which the disposal system, its components and its environment might evolve, termed future lines of evolution. A summary of the future lines of evolution for the disposal system environment and key components is presented in the next section; these future lines of evolution is an important input to Performance Assessment and Formulation of Radionuclide Release Scenarios. The evolution-related FEPs for each component also provide a checklist of processes that should be considered in evaluating the performance of the repository system and its components; this is described in Performance Assessment. This consideration leads to the confirmation that under expected initial conditions and likely future lines of evolution there should be no release of radionuclides over several hundreds of thousands of years, but also identifies the conditions and events that could lead to radionuclide releases. The migration-related FEPs for each component provide a checklist of processes that should be considered in the models for radionuclide release, transport in the repository system, and dispersion and impact in the biosphere, as described in Models and Data for the Repository System and Biosphere Data Basis, Terrain and Ecosystems Development Modelling, Surface and Near-Surface Hydrological Modelling, Biosphere Radionuclide Transport and Dose Assessment and Dose Assessment for Plants and Animals. Thus, those FEPs most important to radiological assessment are carried forward to Assessment of Radionuclide Release Scenarios for the Repository System and Biosphere Assessment. 4.4 Future lines of evolution The understanding of FEPs is used to develop descriptions of the future lines of evolution of the repository system itself (the engineered barriers and host rock) and of the natural setting of the site. This provides the framework for estimating the thermal, hydraulic, mechanical, and chemical loads that will be placed on the system.

137 92 During the construction and operation of the repository up to its closure, the main changes will be related to excavation effects and draining of water from the underground openings, plus introduction of heat from the spent fuel. Rock stress changes and some limited damage immediately around the openings are expected, as well as an increase of the groundwater flow into the repository volume, and hence changes in hydrogeology and hydrogeochemistry. After closure, the groundwater flow regime will return towards preconstruction conditions, although modified by radiogenic heat from the spent fuel for a time. Hydrogeochemical changes induced during the open period may persist for longer because of the very low groundwater flows. In the longer term, the main driver for changes will be climate evolution, where the expected case is a reversion to glacial-interglacial cycling as experienced over the last one million years of the Quaternary. However, best scientific understanding indicates that past and continuing anthropogenic emissions of CO 2 and other greenhouse gases will lead to increased global temperatures over a period of many thousands of years, delaying the onset of cooler climate conditions. Thus, over the next 50,000 years (50 ka), conditions are expected to remain essentially as today, i.e. a temperate climate with the boreal ecosystem. Glaciation-related crustal uplift (isostatic rebound) will continue, although at a decreasing rate, and will outpace global (eustatic) sea-level rise; this will lead to local sea-level fall relative to the land, at least for the next several millennia. Minor changes in hydraulic boundary conditions will occur due to relative sea-level fall and build up of peat deposits in low-lying areas. A first cold period is expected to occur at about 50 thousand years after present (ka AP) with temperature and precipitation changes leading to permafrost development and, later on, to ice-sheet development (see Ch. 4 in Formulation of Radionuclide Release Scenarios based on Pimenoff et al. 2011). The general form and degree of changes beyond this can be estimated based on reconstruction of past global climate changes, although there is uncertainty in the timing of changes. For the assessment, a repetition of the sequence of events during the last glacial cycle (the Eemian-Weichselian) is assumed. Reliable proxy data exist for this cycle, which can be taken as representative. In the assumed forward sequence, the first permafrost period is between 50 and 60 ka AP, with a second permafrost period between about 73 and 81 ka AP, before the onset of three successive periods of ice sheet cover lasting until about 155 ka AP (Figure 4-1). After 170 ka AP, a repetition of the cycle from 50 ka to 170 ka AP is assumed. This is as expected in the absence of anthropogenic effects and with a return to naturally-driven climate cycling. In this case, seven glacial cycles including alternating temperate and cold periods may be expected from 170 ka AP to 1000 ka AP. That is a total of eight glacial cycles in the assessment time frame (up to one million years after present). Realistically, variability between future glacial-interglacial cycles is to be expected (in both duration and intensity), but that this would not substantially alter the results of the performance or safety assessment.

138 93 A ka BP ka AP B Permafrost Ice Sheet Temperate Time window (ka AP) Up Climate type Temperate (T) to 50 T T T Cold /Permafrost (P) P P P P Cold/Ice Sheet (IS) IS IS IS Temperate (T) = climate as today; Permafrost (P); Ice Sheet (IS) Figure 4-1. Schematic representation of the occurrence of permafrost, ice sheets, and temperate safety assessment climate types during (A) the last glacial cycle, (B) assumed repetition of the past glacial cycle from 50 ka AP onwards, and the representation of the future sequence in terms of the climate types in time windows.

139 94

140 95 5 MODELS AND DATA This chapter describes the various models and data needed for the analyses supporting the safety case. Four separate categories of model can be recognised, based on their purpose: models describing the climate evolution and climate-driven processes; these models frame the analysis of the disposal system by giving information on the climate conditions that occur within defined time windows (Section 5.1); models needed to represent the features, events and processes that are the main drivers of the repository system s evolution and used to assess the performance of the engineered barrier system and conditions in the repository host rock (Section 5.2); models used for analysing radionuclide release and transport from the near field through the geosphere to the surface environment (Section 5.3); models used for the biosphere assessment including models describing the development of the surface environment, models describing radionuclide transport in the surface environment and models for assessing potential doses to humans, plants and animals (Section 5.4). Different models are used for each of the components, depending on the processes described. These models are linked, however, such that the initial and boundary conditions as well as the input data for the different models are selected in a consistent way. Thus, the output from one model can be used as an input to the next in the modelling chain. The first three items in the above list are described in full in Models and Data for the Repository System; the last item is described in Biosphere Data Basis and Biosphere Assessment. These reports include references to the studies that describe the models and data in more detail. 5.1 Models and data for climate evolution and climate-driven processes Climate modelling has been carried out to support the definition of the climate scenarios and define the time windows for temperate, permafrost and glacial climate conditions (Pimenoff et al. 2011). The main factors affecting the climate evolution and the onset of the next glaciation are the Earth s orbital variations and associated variations in solar insolation as well as the changing atmospheric CO 2 concentration. Simulations of future climate evolution were made using constant reasonable concentrations of CO 2 and emission scenarios for the current century and the consequent evolution of CO 2 concentrations over the next several millennia. Changes in solar activity and in the concentration of volcanic dusts in the atmosphere may also affect climate, but such changes are uncertain and were not included in the models. The estimation of the climate states on a time scale of 120,000 years is based on analysis of climate simulations of an Earth-System model (CLIMBER-2) coupled with an ice-sheet model (SICOPOLIS). The large-scale output of the climate was downscaled using a regression model (GAM) so that the climate conditions relevant for the Olkiluoto site could be extracted (see Figure 5-1). In addition, a more detailed study of

141 96 the climate evolution on a time scale of 10,000 years was carried out (Pimenoff et al. 2012). In this study, other types of Earth System models (MPI/UW, UVic) were used. The climate modelling approach is summarised in Figure 5-1 and Table 5-1. The modelling results have been used to define the climate scenarios, as well as providing input to the permafrost modelling, groundwater flow modelling and the development of the surface environment at the Olkiluoto site. Permafrost occurs when cold and dry climate conditions prevail for extended periods without ice sheet cover, and beneath cold-based ice sheets. The development of permafrost and frozen ground depends on heat-exchange processes across the atmosphere-ground boundary layers and on an almost time-independent geothermal heat flow from the Earth s interior. A main driver for permafrost development is the evolution of the climate conditions at the site. The development of permafrost at the Olkiluoto site has been modelled by Hartikainen (2013). The model describes the freezing and thawing of groundwater-saturated bedrock either in 1D or 3D by considering heat transfer and freezing accounting for groundwater salinity and hydraulic pressure. The bedrock is considered as an elastic porous medium and the groundwater as an ideal solution of water and ionic solvents. The model has been used to represent the selected future evolution of the climate (see Section 4.4) with the climate conditions based on Pimenoff et al. (2011, 2012). The crustal radiogenic heat production and the heat generated by the spent fuel were taken into account. Key input for the permafrost modelling is shown in Table 5-1. The results of the modelling have been used for the formulation of the radionuclide release scenarios as well as providing input to groundwater flow modelling. The rock responds to the weight of the ice sheet by uplift in front of the advancing ice margin, by depression during the advance of the ice sheet and by uplift during and after the ice sheet retreat. The response of the rock to the ice load is governed by the rheological parameters of the rock. The ice loading has an effect on the rock stresses, hydrogeology and hydrogeochemistry of the site. Post-glacial crustal uplift together with changes in the global sea level is an important factor affecting the development of the surface environment. The post-glacial uplift considered in the biosphere assessment and groundwater flow modelling has been modelled based on the semi-empirical model presented by Påsse (2001), see Table 5-1. The model parameters are defined by fitting the model to relevant shoreline displacement data and by using crustal thickness data. For the current safety case, the parameters are defined (Chapter 9 of the Biosphere Data Basis) taking into account a revision of Bothnian Sea shore-level data and using a derivative method applying crustal thickness and current uplift maps as well as taking into account complementary data e.g. archaeological observations.

142 97 Input to CLIMBER-2 SICOPOLIS Atmospheric CO 2 concentration Solar insolation Atmosphericdust Geothermal heat flux CLIMBER-2-SICOPOLIS Combination of Earth System Model of Intermediate Complexity (EMIC, CLIMBER-2) used for simulating the climate evolution and an Ice sheet model (SICOPOLIS) describing the evolution of the Northern Hemisphere ice sheets Output of CLIMBER-2 Mean near-surface air temperature Mean precipitation Solar flux at the surface Global radiation at the surface Vegetation Output of SICOPOLIS Evolution of NH ice sheets Ice sheets basal temperature Bedrock elevation Mean near-surfacetemperature Mean precipitation Distance to the ice sheet margin Topography GAM Downscaling the output of the large-scale global model to regional scale using a regression model based on regional climate data. Output of GAM Mean near-surface temperature Mean precipitation at Olkiluoto Figure 5-1. Schematic presentation of the climate modelling approach using global models and downscaling the results to the regional scale.

143 98 Table 5-1. Main models and codes used for modelling climate evolution and climatedriven processes for the safety case TURVA Models and codes Climate evolution Purpose and scope Modelling of the climate evolution to support the selection of the climate scenarios and climatic conditions during temperate, permafrost and glacial periods. The key input for the model is shown in Figure 5-1. Main references: Pimenoff et al. (2011, 2012) CLIMBER-2 Earth System Model of Intermediate Complexity (EMIC) used for simulating the climate evolution (near-surface air temperature, precipitation, solar flux at the surface, global radiation at the surface and vegetation) on a time scale of 120,000 years. SICOPOLIS Ice-sheet model describing the evolution of the Northern Hemisphere ice sheets, their thickness and areal extent, basal temperature and bedrock elevation. GAM Generalised Additive Model used to downscale near-surface air temperature and precipitation from the CLIMBER-2-SICOPOLIS results to the Olkiluoto area. MPI/UW Earth System Model of Max Planck Institute used for the estimation of the climate evolution (atmospheric CO 2 concentration, near-surface air temperature, precipitation, sea level, incoming shortwave radiation at the surface) on a time scale of 10,000 years. The model enables a coupled treatment of atmosphere, ocean, sea-ice, ocean carbon cycle and dynamic vegetation. UVic Earth System Model of the University of Victoria used for the estimation of the climate evolution (atmospheric CO 2 concentration, near-surface air temperature, sea level, incoming shortwave radiation at the surface) on a time scale of 10,000 years. The model enables a coupled treatment of atmosphere, ocean, sea-ice, ocean carbon cycle and land and terrestrial vegetation carbon cycle and oxic-only sediment respiration. Permafrost modelling Modelling of permafrost development e.g. depth of the permafrost to support the formulation of the radionuclide release scenarios as well as input to groundwater flow modelling. Key inputs to the permafrost modelling are the climate conditions, e.g. air temperature and vegetation, based on the climate model (see above), soil cover and water bodies and the properties of the rock mass and groundwater based on Olkiluoto specific data and the heat generated by the spent fuel. Main reference: Hartikainen (2013). 1D model Freezing and thawing of groundwater saturated rock by applying a 1Dmodel. 3D model Freezing and thawing of groundwater saturated rock by applying a 3Dmodel. Crustal uplift Crustal uplift for groundwater flow modelling and the development of the surface environment. Key inputs for the model are shoreline displacement data and crustal thickness data. Main references: Påsse (2001) and Chapter 9 of Biosphere Data Basis

144 Key models and data for performance assessment and for formulation of radionuclide release scenarios The aim of performance assessment is to evaluate the behaviour of the engineered barrier system and, in particular, to confirm that the performance targets will be met and that the host rock conditions will remain consistent with the target properties under the different conditions of the repository system evolution (operational period, thermal period and repository saturation, temperate climate, permafrost and glaciation). The main modelling activities (Figure 5-2) in support of the performance assessment are: modelling of the hydraulic evolution, hydrogeochemical evolution and rock mechanical evolution of the geosphere giving estimates of flow rates, flow paths, groundwater composition, stress state, rock damage around the underground openings including deposition tunnels and deposition holes and shear displacements along fractures; description of the evolution of the closure, which considers degradation of the closure with consequent release of alkaline leachates; modelling of the mechanical and hydraulic evolution, geochemical evolution and mechanical and chemical erosion of the buffer and backfill, which results in an estimate of the mass loss, porewater composition and hydraulic conductivity of the buffer; modelling of the initial defects of the canister, the canister corrosion, mechanical evolution of the canister and creep, which results in an estimate of the overall canister performance; thermal evolution is considered in the modelling of each of the barriers. The models applied consider the main processes affecting the performance of a barrier. There are couplings between the processes and across the barriers. These are not always modelled explicitly, but are taken into account by selecting consistent boundary and initial conditions. The models and data used for assessing performance of the disposal facility are discussed in Models and Data for the Repository System and the results summarised in Performance Assessment. Key input for the assessment is the initial state of the repository system and the barriers provided by the Production Line reports and summarised in Description of the Disposal System. A summary of the models and codes used in support of the assessment of the geosphere evolution is presented in Table 5-2 and of the EBS in Table 5-3.

145 100 Figure 5-2. Overview of the models used for performance assessment. Permafrost depth and ice-sheet thickness from climate and climate-driven modelling (Section 5.1) and results from surface environment modelling (Section 5.4) are applied to assessment of the performance of the repository system. Green boxes present the main processes modelled and yellow ovals present the main outcomes of the models.

146 Models and data for geosphere evolution Modelling of geosphere evolution hydraulic evolution, geochemical evolution and rock mechanical evolution aims to determine whether favourable conditions, such as low flow rate, reducing and also otherwise favourable groundwater chemistry, and mechanically stable characteristics prevail at the repository depth. The main modelling activities and codes used for describing the geosphere evolution are presented in Table 5-2. The modelling assesses the impact of the excavation of the tunnels, presence of open tunnels, heat generated by the spent fuel, as well as of the natural phenomena related to the ongoing crustal uplift and to future glaciations. Site Description provides the description of the bedrock and the groundwater system at the Olkiluoto site, and, together with Features, Events and Processes, the interacting processes and mechanisms. Site Description also provides input to modelling the geosphere evolution. The modelling of geosphere evolution, in turn, provides input for assessing the performance of the EBS. Groundwater flow modelling has been carried out to represent the hydraulic evolution and to assess the flow conditions including flow rates, flow paths both to and from the repository, and the salinity evolution in the geosphere, especially around and through the repository and underground facilities. Two types of flow models, equivalent continuous porous medium (ECPM) combined with a dual porosity (DP) approach and discrete fracture network (DFN) models, have been applied in the groundwater flow modelling. The ECPM conceptualisation has been used mainly to simulate the evolution of the groundwater flow at the site scale. The site-scale model modelling considers density driven groundwater flow and thermal conduction in order to study the effect of the heat generated by the spent fuel and the effect of variable salinity. In the DP approach, advection and dispersion are the dominant processes within the water-bearing fractures, whereas in the rock matrix, solutes are transported by diffusion. DFN models have been used to describe the distribution of the groundwater flow on a detailed scale in the vicinity of deposition tunnels and deposition holes. The approach is based on a stochastic representation of the bedrock fractures. The main processes considered in the geochemical evolution in the bedrock are mixing of groundwaters, and water-rock interactions. The hydrogeochemical evolution at the site under the temperate climate during the first 10,000 years and during ice-sheet retreat has been studied by reactive transport modelling applying FASTREACT (FrAmework for Stochastic REACtive Transport, Trinchero et al. 2010). The model evaluates the mixing of the infiltrating waters with the initial waters, taking into account the main reactions between these mixed waters and rock matrix and fracture minerals along the streamlines. Microbial populations and processes also affect the groundwater composition. The microbially mediated reactions have, so far, not been explicitly taken into account in the reactive transport modelling. However, considering only inorganic oxygen consumption in the reactive transport modelling is cautious as the microbial reactions would increase oxygen consumption. The role of microbially mediated reactions in sulphate reduction has been assessed based on the extensive groundwater and microbial sampling data from the site. These reactions have an important role in defining the redox conditions in the host rock. Leachates from the cement components

147 102 e.g. grouts and in plugs, will also affect the groundwater composition (see discussion in Geochemical evolution of backfill and buffer in Section 5.2.2). The main modelling activities related to rock mechanical evolution have focussed on the disturbances caused by the excavation and the thermal load generated by the spent fuel, stress evolution and rock stability during a glacial cycle, and the rock shear displacements caused by future earthquakes. Different methods have been applied to estimation of the spalling around the underground openings including deposition tunnels and deposition holes; traditional continuum thermomechanics and fracture mechanics approaches as well as analytical methods. Statistical methods have been used for defining the input data. The evolution of the rock stresses during the different phases of a glacial cycle has been described. Further, the interaction of the in situ stress state and the fault zones, i.e. how the fault zones affect the stress magnitudes and orientations, and the stability of the faults under different stress states have been studied. Shear displacements in fractures induced by post-glacial seismic events in nearby fault zones have also been studied. The modelling considered a selection of the fault zones at Olkiluoto with varying orientation, extent and location with respect to the repository. Table 5-2. Main models and codes used for host rock evolution in TURVA Models and Purpose and use codes Hydrological evolution Groundwater flow modelling has been carried out to describe the groundwater flow rates, flow paths and salinity distribution in the geosphere and around the repository. Two conceptualisations of the flow have been applied, equivalent continuous porous medium (ECPM) combined with a dual porosity (DP) approach and discrete fracture network (DFN) models. The shore line evolution and the assumed climate conditions are taken into account by initial and boundary conditions and the repository layout and the underground openings and their properties are represented in the models. The results are applied to assessing the geochemical evolution and performance of the EBS, and they also provide input to the radionuclide release and transport analysis. Main references: Löfman & Karvonen (2012), Hartley et al. (2013 a,b,c) ConnectFlow Software package for simulation of the groundwater flow at different scales using both ECPM and DFN approaches and their combination. Enables detailed study of the flow paths by the stochastic representation of individual fractures and representation of the underground openings in the model. The hydrogeological DFN models and hydrogeological structure model are key inputs to the model. Three different scales (see Figure 5-3); regional, site and repository, with varying levels of detail of the representation of the sparsely fractured rock, hydrogeological zones and underground openings (e.g. deposition holes, deposition tunnels, other tunnels and shafts) are used. Main results used in the safety case are the inflow to and flow rates around the deposition holes and deposition tunnels, flow paths between the repository and the surface environment and flow and transport properties along these paths, but the salinity evolution has also been modelled. FEFTRA The finite-element program package for groundwater flow modelling which applies the ECPM approach to model transient and density-driven flow and heat transfer by conduction and the DP approach for modelling salt transport. The rock is represented by two hydraulic units, hydrogeological zones and sparsely fractured rock. Both of these units have averaged hydraulic properties based on site-specific data. The tunnel system is considered by using the appropriate boundary conditions to represent the hydraulic or thermal impact of tunnels. Key inputs to the model include outputs from the hydrogeological structure model and the hydraulic conductivity of the sparsely fractured rock, properties of the rock related to salt transport and thermal properties of the rock. Main results used in the safety case are the evolution of the flow conditions and groundwater salinity. Geochemical evolution

148 103 Models and codes Purpose and use Assessment of the geochemical evolution of the site is based on understanding of the past evolution of the site as well as on reactive transport modelling. The main processes considered are mixing of groundwaters, and water-rock interactions. The FASTREACT (FrAmework for Stochastic REACtive Transport, Trinchero et al. 2010) approach has been used to combine the flow and the chemical reactions. In this approach, particle tracking methods are used to define particle trajectories, streamlines, along which reactive transport simulations are carried out. Expected and bounding groundwaters for different future conditions are defined based on these assessments and are used for assessing the performance of the EBS as well as for determining the solubility, speciation and retention parameters for the radionuclide release and transport analysis. Main references: Models and Data for the Repository System, Trinchero et al. (2013), Wersin et al. (2013c) PHREEQC Reactive transport modelling code used for assessing the evolution of the groundwater chemistry at the site. The solute concentration in the whole set of particle trajectories is reproduced using a set of PHREEQC one-dimensional reactive transport simulations where the longitudinal coordinate (e.g. the distance from the infiltration location) along the trajectory is interpreted in terms of travel time. The key inputs to the model are the velocity field and the initial salinity field derived from the FEFTRA model for representative times, which are used to define the streamlines. Further inputs are site-specific mineralogical data and the composition of the initial and infiltrating waters. The main results are the groundwater composition at the repository depth and evaluation of the buffering capacity of the host rock against the infiltrating waters. Rock mechanics evolution Modelling has been carried out to estimate the disturbances caused by the excavation and thermal load generated by the spent fuel, the stress evolution at site during a glacial cycle and the rock shear displacements caused by earthquakes. The results have been taken into account in the groundwater flow modelling and in estimation of the likelihood of canister failures due to shear displacements. Main references: Site Description, Ch. 9.2, Valli et al. (2011), Lund & Schmidt (2011), Fälth & Hökmark (2011, 2012), Hakala et al. (2008) 3DEC A rock mechanics code, a three-dimensional numerical program based on distinct element method for discontinuum modelling. 3DEC simulates the response of the (such as jointed rock mass) subjected to either static or dynamic loading. Thermal loads can be taken into account. The discontinuous medium is represented by an assemblage of discrete blocks and the discontinuities are treated as boundary conditions between the blocks. The code has been used for several studies including spalling predictions and assessment of the effect of the faults on the in situ stress field and analysis of the shear displacements due to end-glacial earthquakes. Key inputs for the analyses have been site-specific data on rock mechanics properties of the rock mass, fractures and deformation zones and thermal properties of the rock, rock stresses and the estimated effect of the glacial load. Fracod2D A fracture mechanics code based on the Displacement Discontinuity Method (DDM) that has been used for predicting potential for spalling. Key input for the analysis have been sitespecific data on rock mechanics properties of the rock mass, fractures and deformation zones and thermal properties of the rock and rock stresses. ABAQUS The finite element code used for calculating the glacially induced stresses, which combined with a synthetic regional background stress model, is used for assessment of the fault stability. In addition to the stress model, key inputs for the analysis include a model of the Weichselian ice sheet, and variant Earth models for the lithosphere.

149 104 Figure 5-3. Groundwater flow models and data flows between them (after Hartley et al. 2013b) Models and data for engineered barrier system performance The models and data used for assessing performance of the EBS are discussed in Models and Data for the Repository System and the results summarised in Performance Assessment. This section summarises the main models, codes and data used (for a summary, see Table 5-3). Thermal evolution The model that encompasses both the EBS and the geosphere is the thermal evolution model. Decay heat from the spent nuclear fuel will increase temperatures within and around the repository for up to 50,000 years. In developing the repository design and layout, the principal constraint is to limit the maximum temperature experienced at the buffer/canister interface to 100 C; this is to preclude significant thermal alteration that might degrade the desired swelling and hydraulic properties of the buffer. The temperature evolution also affects groundwater flow and permafrost development. Therefore, the thermal evolution needs to be modelled for the entire repository system. Heat is transferred from the canister through the host rock mainly by conduction through the different materials between the canister and the wall rock. The conductivity of the copper, of the bentonite buffer (and backfill), the degree of saturation of the buffer and the presence of air/water gaps at the interface with the canister and with the rock also affect the near-field thermal evolution. Heat is transferred through the

150 105 geosphere by a combination of conduction in the solid rock mass and advection of flowing groundwater through fractures. The thermal properties of the rock/groundwater system, together with groundwater flow, will therefore govern the temperature evolution throughout the geosphere. Diffusivity (of heat) is the most important thermal parameter, which is a function of thermal conductivity, heat capacity and density of the rock mass. Thermal dimensioning of the repository and the resulting temperatures are presented in Ikonen & Raiko (2013). The model is described in Models and Data for the Repository System. Mechanical, hydraulic and geochemical evolution of closure components There is no specific performance assessment model for the mechanical, hydraulic and geochemical evolution of closure components. Degradation of cement and the various processes involved are qualitatively discussed in Section of Performance Assessment. In the groundwater flow model, the decreased performance of the deposition tunnel and closure backfill is taken into account (Hartley et al. 2013b) assuming that the hydraulic conductivity of the tunnels is one to two orders of magnitude higher than the reference assumptions. Based on experimental data and monitoring of leachates of cement grouting of fractures in ONKALO, the evolution of ph and alkali contents in cement leachates has been evaluated. The output from this estimate is then used to assess the impact of cement leachates on buffer and backfill in Performance Assessment. Mechanical and hydraulic evolution of backfill and buffer A model for the mechanical and hydraulic evolution of the backfill and buffer is needed to evaluate the capability of the buffer and the backfill: to protect the canister from rock movement, to provide a favourable hydraulic and mechanical environment for the canister, to limit transport of harmful substances (e.g. sulphides) to the canister, and to limit transport of radionuclides out of the canister in case of a release. The main relevant properties of the buffer and the backfill are swelling pressure and hydraulic conductivity. These properties depend, among other parameters, on the thermohydraulic behaviour of the clay. The finite element code CODE BRIGHT is used to model the thermo-hydraulic behaviour of clay. Although the code is able to represent the mechanical behaviour of a porous medium in a coupled form, only the thermal and water flow capacities of the code have been used. In CODE BRIGHT, equations for mass balance were established following the compositional approach. That is, mass balances were performed for water and air instead of using solid, liquid and gas phases. The equation for balance of energy was established for the medium as a whole. The final objective was to estimate the temperature, T, and liquid pressure, P l, for the thermo-hydraulic analysis from water mass and energy balance equations. Gas pressure was assumed constant in these analyses.

151 106 The mechanical model reproduces the behaviour of the porous medium under mechanical boundary conditions (displacements and forces or stresses). This is called a constitutive model and for this application the model used is the BBM (Barcelona Basic Model). The model is formulated within the framework of hardening plasticity using two independent sets of variables: the excess of stress over air pressure and suction. The BBM model is able to represent many of the fundamental features of the behaviour of partially saturated soils. On reaching saturation, the model becomes a conventional critical state model (Schofield & Wroth 1968). Geochemical evolution of backfill and buffer The following processes potentially affecting the main properties of the buffer, such as its swelling pressure and its transport properties. These are assessed either through modelling or by reference to experimental data. Oxygen depletion and changes in ph: The end point is the time it takes for all the originally trapped air to be consumed in the buffer and in the backfill, i.e. how long it takes for the conditions to become anoxic in the near field. This is estimated using a mass-balance approach or a semi-empirical approach that takes into account the reaction rate of O 2 with pyrite in saturated or unsaturated conditions (see Section in the Performance Assessment for details). Evolution of colloid population: The evolution of the population of colloids introduced with the buffer and the backfill, especially those associated with the degradation of repository materials, and their mobility and stability under changing groundwater conditions is not yet well defined. Bentonite colloids in low ionic strength groundwaters are currently being addressed in the CFM project in Grimsel (Möri 2004) and in the Äspö Colloid Project (Laaksoharju & Wold 2005). There are also some experimental data from URL studies to date although in the Äspö Colloid Project, bentonite colloids were shown to be unstable in the high ionic strength deep Äspö and Olkiluoto groundwaters, and so the concentrations were as low as those in the natural groundwater. In the Performance Assessment the initial colloid concentration was neglected (Section in the Performance Assessment) Effect of cementitious leachates on the buffer and backfill: The impact of cementitious leachates on the engineered barrier system and the rock is assessed based on experimental data and monitoring of leachates from cement grouting of fractures in ONKALO (Sections and in Performance Assessment). The effects of leaching of other sealing materials (Silica sol) on bentonite are considered through experimental data (Section in Performance Assessment). Geochemical evolution of unsaturated buffer porewater during the thermal stage: This has been assessed by thermo-hydro-geochemical modeling, using the integral finite-difference code TOUGHREACT. The objective of the modelling was to identify changes in porewater chemistry and to assess the redistribution of salts and minerals that could induce cementation (Section in Performance Assessment). Buffer and backfill porewater chemistry after saturation: The porewater chemistry and its evolution have been modelled by coupled diffusion-reaction modelling, accounting for the evolving groundwater composition. The porewater composition is approximated assuming diffusive equilibration and chemical equilibrium between

152 107 the groundwater and the clay. Porewater composition has been modelled using the code PHREEQC. This has been done in a thermodynamic model for bentonite, which is based on the microstructure, the electrochemical properties of the clay and the anion exclusion concept (Wersin et al. 2013a) (Section in Performance Assessment). Microbial activity in the buffer and sulphide production at the buffer/rock interface: The gypsum (sulphate) pool in the buffer will dissolve, and may be a source of sulphide production at the buffer/rock interface in zones of larger porosity through the activity of sulphate-reducing bacteria (SRB). This process was considered in a bounding analysis, using both an analytical shrinking core model and a linear 1D reactive transport model. From these considerations, maximum sulphide fluxes towards the canister were estimated (Section in Performance Assessment). Microbial activity in the backfill and the effect of organic materials on sulphide production in backfill: No process-based model has yet been developed. Microbially induced sulphide production is evaluated through experimental data, a mass balance approach and chemical kinetics considerations (Sections and in Performance Assessment). Microbial activity, organic carbon and sulphide production in backfill: Gypsum and organic carbon in the backfill are potential sources for microbial activity and sulphide production, which might occur in low porosity zones at the backfill/rock boundary. This rather complex process was addressed with a step-wise approach. First, a bounding analysis including an analytical shrinking core model and a 1D reactive transport model was conducted to estimate maximum sulphate and sulphide fluxes into the rock and the buffer (Wersin et al. 2013c). Then, iron sulphide formation, an important sink for sulphide, was addressed by simple equilibrium calculations and natural analogue data. Finally, a more advanced, but still preliminary coupled reactive transport model with different boundary assumptions at the backfill/rock interface was applied to verify the results obtained from the simple model considerations, to assess both sulphur and iron fluxes and to evaluate uncertainties (Sections 6.6.3, 6.5.7, and in Performance Assessment). Montmorillonite transformation: the processes of illitisation and cementation have been assessed using semi-empirical approaches and data from natural analogues. For illitisation, two approaches were used: kinetic rate equations as a function of the temperature and mass transfer constraints. Natural analogues considerations were used to assess the smectite-to-illite conversion rates for relevant repository conditions (see Section of Performance Assessment). Cementation was assessed through experimental data from the LOT experiment (Sena et al. 2010, Karnland et al. 2009) and reactive transport modelling to calculate the amount of silica precipitated from dissolution-precipitation processes of accessory minerals (see Section of Performance Assessment for details). For the long-term performance of bentonite, thermodynamic and kinetic considerations as well as natural analogue evidence were used (Section of Performance Assessment). The effect of the canister corrosion products on buffer stability was assessed on the basis of empirical data from the LOT experiment (Section of Performance Assessment). Iron-clay interactions in the backfill are discussed using a mass balance approach taking into account the rock bolts and the remaining steel mesh

153 108 and metal debris which are in direct contact with the backfill. The results are discussed in Section of Performance Assessment. Mechanical and chemical erosion of buffer and backfill Mechanical and chemical erosion of the buffer and the backfill are processes that may lead to mass loss both with consequent loss of swelling pressure and an increase of hydraulic conductivity. Mass loss in the buffer can also affect the ability to filter colloids in the buffer and to limit microbial activity. Therefore, the processes that may lead to erosion have to be modelled to assess the effect on the transport properties of the buffer and the backfill. For mechanical erosion, the background data and basis for mass loss estimation are presented in Section 5.4 of Performance Assessment. A simplified approach is used to estimate buffer and backfill mass loss. This approach is based on the calculation of mass loss as a function of the total volume of water flowing through a piping channel in the buffer or backfill and the erosion rate. The volume of eroding water is based on the distribution of the groundwater flow between the deposition holes and the deposition tunnels and thus it depends on the groundwater flow model results. The erosion rate is based on experimental tests done by Posiva and SKB in conditions relevant to those at Olkiluoto in terms of salinity, flow and geometric properties of the fracture potentially intersecting the deposition tunnel or deposition hole. The eroded mass loss is derived by numerical integration of the measured erosion rate as a function of cumulative flow of water. From the eroded mass loss, the impact on hydraulic conductivity and swelling pressure can be estimated. The chemical erosion model for predicting the rate of erosion of the bentonite buffer in low ionic strength water used has been developed by Moreno et al. (2010). The model details of chemical and surface chemical processes related to chemical erosion, the input data and experimental results are presented in Section of Performance Assessment. The model uses the results of discrete fracture network (DFN) flow modelling to estimate the chemical erosion rate. Three processes are modelled: 1) Transport of sodium ions in montmorillonite pore water, 2) Expansion of montmorillonite in the fracture and 3) Flow of montmorillonite gel and sol and of water. The same model has been used to make equivalent calculations for the backfill by Sane et al. (2013). Schatz et al. (2013) performed a series of small-scale, flow-through, artificial fracture experiments in which swelling clay material could extrude/erode into a well-defined system representing a fracture, in order to provide the basis for modelling the potential extrusion/erosion behaviour of bentonite buffer material at a transmissive fracture interface. The chemical erosion model was applied to the repository at Olkiluoto, using the following data: the evolution of groundwater velocity with time in fractures intersecting the deposition holes and tunnels; the transport apertures of these fractures;

154 109 the periods during which low-ionic strength conditions can be expected to prevail around each deposition hole. A key input assumption is the amount of buffer and backfill mass loss that would lead to advective conditions in the deposition holes. The data used are from Section of SR-Site (SKB 2011). The end point of the model is the number of deposition holes in which advective conditions are established as a function of time based on the mass of eroded buffer (or backfill). The output of this model is then used as input for the canister corrosion calculations by sulphide attack in the long term (see canister corrosion, below). Freezing and thawing of buffer and backfill Freezing of the buffer or backfill is not expected as permafrost is not expected to reach repository depths. If it occurred, freezing of the bentonite porewater and/or backfill porewater would be expected to affect swelling pressure. A theoretical description of the temperature dependence of swelling pressure for the buffer was derived (Birgersson et al. 2010). The pressure responses of fully saturated buffer and backfill material samples have been observed down to -10 C. The results are discussed in Section 7.3 of the Performance Assessment. Canister corrosion The evolution of the copper overpack has been assessed. The following corrosion processes are considered: atmospheric corrosion, localised corrosion, generalised corrosion due to sulphide (including microbially induced corrosion) and to oxygen, and stress corrosion cracking as well as copper corrosion in oxygen-free water. The corrosion loads were determined by assessing the total amount or the flow of a given corroding agent toward the copper surface and the duration of a given corrosion process. For each of the identified corrosion processes and corrosion loads, calculations were carried out to estimate the corrosion depth that the processes can induce on the copper wall. The extent of damage on the canister surface due to localised corrosion, mostly from chloride ions, has been assessed from experimental and natural analogue results rather than by calculation. A given corrosion depth (rather than using a pitting factor, as was done in the past) is then attributed to the effect of localised corrosion from chloride ions. The generalised corrosion of the copper canister due to oxygen and sulphide has been calculated based either on mass balance approach or on diffusive flow of the corroding agent to the copper surface. The flow of corroding agent (e.g. oxygen or sulphide) assumed to react with the canister is estimated using the results from the geochemical models for groundwater and for the buffer and backfill (see above). The groundwater flow conditions in each deposition hole (derived from the DFN model) are taken into account to determine the flow of corroding agent to the canister.

155 110 The state of the buffer with respect of erosion is another important assumption affecting the diffusion rate of corroding agents to the copper surface. The copper surface that is exposed to the corroding agent (i.e. the entire surface of the canister or a limited area of the canister) determines the corrosion depth and ultimately the durability of the canister. In the case of stress corrosion cracking and other postulated corrosion processes (e.g. copper corrosion in anoxic water) the conditions in which they occur are discussed in Features, Events and Processes and reported in Performance Assessment. The corrosion depths expected from the different corrosion processes during the near-field evolution are then compared with the total corrosion depth and the number and timing of canister failures by corrosion are thus estimated. Copper creep and copper overpack lifetime The stresses and strains developing in the copper canister during its first decades of repository service are modelled using finite element analysis simulations optimised to predict the EB-welded copper canister creep life. The creep model takes into account the swelling of bentonite around the canister and the presence of initial residual stresses. The model includes elastic-plastic as well as creep simulations. The model simulates the mechanical properties of the base material and that of the EB weld and it includes the base material creep rupture model, base material creep strain model, as well as the EB weld strain model. The uniaxial creep model is translated to a multiaxial constitutive equation form using a VTT in-house translation routine developed for the use of creep models that are not directly supported by the computational tool ABAQUS (Holmström et al. 2013). Mechanical loads on the canister The mechanical loads on the canister have been assessed for the design of the canister so that the canister fulfils its containment safety function for several hundred thousand years. The expected mechanical loads during canister evolution are due to the following: swelling of the bentonite (even or uneven), isostatic load due to the presence of an ice sheet on the surface and dynamic loads in the case of a rock shear movement that could happen in connection with the advance or retreat of an ice sheet. The mechanical response of the canister is analysed using 2D- or global 3D-finiteelement models, including large-deformation and non-linear material modelling as described in Raiko (2012) and in Raiko et al. (2010). In some cases, the creep model developed by SKB has been used, as described in Raiko et al. (2010). In the case of rock shear movement, the mechanical response of the canister as well as that of the buffer surrounding it is analysed. In addition to the expected loads, a disturbance scenario of freezing of the bentonite buffer down to -5 C during permafrost has been analysed using the same model (Raiko 2012, Section 8.4.4). The canister mechanical integrity is assessed partly from the stress and strain results obtained using global models and partly from fracture resistance analyses using the sub-modelling technique (see below). The sub-model analyses utilise the deformations from the global analyses as constraints on the sub-model boundaries

156 111 and more detailed finite-element meshes are defined with defects included in the models, together with elastic-plastic material models. The canister mechanical analyses are carried out in collaboration between SKB and Posiva (as described in Raiko et al. 2010). The SKB data are considered applicable to Olkiluoto because of the similarities in future climatic conditions. In the case of the isostatic load from the ice sheet, the isostatic load used by SKB is more severe than the one expected at the Olkiluoto site since, according to the climate model (see Section 5.1). The maximum thickness of the ice sheet expected at the Finnish site is 2 km whereas that expected at the Swedish site is 3 km. Number of initial penetrated canisters The potential number of defective canisters that might be emplaced in the repository has been analysed probabilistically by Holmberg & Kuusela (2011). The main objective of this study was to estimate the reliability of the welding and of the non-destructive testing (NDT) processes for detecting penetrating defects in the copper overpack. After varying the assumptions used as input data, it was concluded that the currently available data are insufficient to make a reasonable estimate of the probability of emplacing a penetrated canister in the repository. A Bayesian approach was therefore used based on the existing experience on welding and NDT method reliability. The outcome of the model is a preliminary estimate of the number of initially defective canisters that may be incidentally emplaced in the repository. Table 5-3. Main models and codes used for EBS performance assessment in safety case TURVA Models and codes Purpose and use Thermal evolution Modelling activity carried out to describe the evolution of the temperature in the near field (at the canister surface in the buffer and at the buffer/rock interface) and in the far field. Analytical and numerical approaches developed by VTT are used. Main references: Ikonen & Raiko (2013), Raiko (2012). Analytical and numerical approaches Analytical and numerical approaches are used to model the heat transfer among the different barriers. The processes represented are conduction, radiation and convection. The input data are principally the heat output of the canister (which in turn is determined by the spent nuclear fuel loading and the storage time of the spent nuclear fuel prior to emplacement in the canister), the repository layout and the thermal properties of the canister, bentonite buffer and backfill, and the surrounding rock. Mechanical, hydraulic and geochemical evolution of closure components No codes, empirical data only. Main references: See Sections 5.6, 6.7 of Performance Assessment. Mechanical and hydraulic evolution of backfill and buffer Modelling activity carried out to describe the duration of saturation and piping and erosion issues. Main references: Sections 5.4 and 6.4 of Performance Assessment. CODE BRIGHT (Olivella et al.1994, 1996) The finite element code CODE BRIGHT is used to model the thermohydraulic behaviour of clay. Although the code allows study of the mechanical behaviour of a porous medium in a coupled form, only the thermal and water flow capacities of the code have been considered. Initially, the code was developed for non-isothermal multiphase flow of brine and gas through porous deformable saline media. Key assumptions used in the material models include: Fourier s law for heat transport, Fick s law for non-

157 112 Models and codes BBM (Barcelona Basic Model, Alonso et al. 1990) Purpose and use advective liquid flow (diffusion) and Darcy s law for advective liquid flow. The heat flow from the canister and the temperature and liquid evolution at the boundaries are the key boundary conditions. The initial conditions are the initial temperature and the liquid pressure distribution, which is considered hydrostatic in rock and negative in buffer and backfill (they are unsaturated at the initial state). The BBM is a critical state model that reproduces the mechanical behaviour of unsaturated soils under different boundary conditions (displacements and forces or stresses). The BBM is an extension of the modified Cam clay model that has become popular in applications involving unsaturated soils and, in particular, in simulations using the finite element method. Partially saturated soils can be loaded in different ways, for instance, mechanically and/or hydraulically. In addition, cycles of loading and unloading can be applied. Key input data used in the model include: the initial porosity, initial temperature and initial liquid pressure, which is negative because the materials analysed are unsaturated at the initial state. The initial stresses are used as well. The boundary conditions in the mechanical model are stresses or constant rates of displacement (usually zero, i.e. fixed boundary). Geochemical evolution of backfill and buffer The following are the main codes used for performance assessment (processes assessed only through data are discussed in the text): PHREEQC Code used to describe the buffer and backfill porewater composition based on a thermodynamic equilibrium model (described in Wersin et al. 2013a, 2012b). The key input data for the buffer are the groundwater composition and the mineralogical and pore water composition of the bentonite. The dissolution and precipitation of accessory minerals as well as surface reactions including cation exchange and protonation/deprotonation are taken into account. TOUGHREACT Integral finite difference code used for thermo-hydro-geochemical modelling (Xu et al. 2008, Idiart et al. 2013). The objective of the modelling is to identify changes in porewater chemistry during the early evolution of the system. In addition to mineral reactions, cation exchange was accounted for in the geochemical model. Key input includes mineral composition of the buffer and backfill, groundwater composition and groundwater flow rates in the repository near field. Mechanical loads on canister 2D- or global 3D-finite-element models are used including large-deformation and non-linear material modelling and, in some cases, also creep. The models include material models for copper and bentonite mechanical behaviour. All materials are modelled according to the elastic-plastic material model developed by von Mises (von Mises 1913). The material models include strain hardening, in some cases also strain rate hardening, swelling pressure dependency, and temperature dependency. In elevated temperature analyses, creep models are also used for copper and in some cases also for iron. Codes applying the finite element method (FEM) are also used to simulate material flaws or lack of material. In the case of cracks, fracture parameters are calculated and the allowable crack sizes are determined. The bentonite material model is described in Börgesson et al. (2010). The swelling pressure and the yield strength of the saturated bentonite are strongly dependent on the density of the bentonite. Different strength and swelling pressure estimates for Na and Ca bentonites are used. The rock shear analysis used a strain-rate dependent material model, so the material stress-strain curve was presented for static and dynamic strain rates (Dillström 2010, Hernelind 2010). Main references: Raiko (2012) and Raiko et al. (2010). ABAQUS (Hillbit et al. 1994) and ANSYS ( Codes used in 3D FEM-analyses. ABAQUS is used to model the bentonite material and the rock shear response of the buffer/canister/insert system. ANSYS is used to simulate static (stationary), dynamic (moving) and heat transfer (thermal) problems. Key inputs to the codes include the material models described above as well the loading on the canister buffer system e.g. isostatic load, the shear load in terms of velocity and forced displacement. The geometry of the canister and buffer is modelled in full 3D with solid elements, including gaps and contacts, and fillets of structural corners.

158 113 Models and codes Purpose and use Copper creep and copper overpack lifetime The creep model used for canister overpack creep simulation has been developed for Posiva by Holmström & Auerkari (2006). The model is based on the logistic creep strain prediction model, which is a creep strain prediction tool able to predict representative creep strain curves and strain rates in a large stress and temperature range. The geometry used in the simulations is that of the OL1 and OL2 (BWR) spent fuel canister. The creep model developed by Sandström et al. (2009) has been used in the canister design strength analyses carried out in collaboration with SKB. Number of canisters with an initial penetrating defect Probabilistic calculation to assess the probability to emplace one or more canisters with an initial penetrating defect in the repository. Main reference: Holmberg & Kuusela (2011) Estimation of the number of canisters with an initial defect based on the Bayesian method Combines information about the reliability of the welding process (Ronneteg et al. 2006, p. 105) and that of the NDT methods as well as human error (Swain & Guttmann 1983) to assess the number of canisters with an initial penetrating defect that could be accidentally emplaced in the repository. Key input assumptions are the reliability of the sealing process (only one of 100 canisters might have a critical flaw) based on the opinion of manufacturing experts and the probability for human error during the NDT process (0.003) based on a general screening value for human errors in nuclear power plant risk analysis. 5.3 Models and data for the analysis of radionuclide release scenarios In the models used in the analysis of most release scenarios, radionuclides released from a failed canister are dissolved in water and conveyed in solution through the repository near field and through the geosphere towards the biosphere (gas- and colloid-mediated transport are also considered in some calculation cases). It is assumed that radionuclides migrate from the repository near field to the geosphere, but not vice versa. Similarly, radionuclides may migrate only from the geosphere to the biosphere and not vice versa. These model simplifications are cautious and consistent with the assumption that radionuclide transport in the geosphere is dominated by advection (retarded by matrix diffusion and sorption) from the near field to the biosphere. The modelling of radionuclide release, retention and transport in the repository system is carried out in two steps: near-field release, retention and transport modelling, and geosphere retention and transport modelling. The near field comprises the spent fuel, the deposition holes including the canisters and buffer, the backfilled deposition tunnels and that part of the immediately surrounding host rock that is affected by the presence of the repository (e.g. excavation-disturbed zones). The geosphere comprises the remainder of the host rock. The models and flow of information in near-field and geosphere modelling are shown in Figure 5-4 and the codes used are shown in Figure 5-5 and summarised in Table 5-4. The models, flow of information and the codes used in biosphere assessment including dose assessment are also shown in these figures, and are described in more detail in Section 5.4. To ensure the traceability of the modelling calculations, a procedure has been developed for the management and documentation of the input and output of the various models. This procedure has been applied to each calculation case; it includes the specification of the models and parameter values to be used, checking of output files, some of which are

159 114 used as input files in subsequent steps in the calculation chain, and the storing of both input and output in the assessment database. Bedrock description Description of fuel and engineered barrier system Near-field flows Near-field release and transport Groundwater flow modelling Geosphere transport resistances Geosphere paths and release locations Radionuclide transport in the geosphere Description of surface environment Terrain and ecosystem development model Landscape model set-up Surface and near-surface hydrology model Yes Dose calculations needed? Annual activity release Landscape model (radionuclide transport in the biosphere) Complementary indicators Biosphere assessment Dose models Annual doses to humans Absorbed dose rates to plants and animals Assessment endpoints Figure 5-4. Models and information flows. Radionuclide release and transport models are shown in white boxes. System descriptions and understanding are shown in light blue boxes, key supporting models in green boxes and their principal outputs in dark blue ovals.

160 115 Near field GoldSim Geosphere MARFA or GoldSim (for PSA) Dose assessment needed? No Annual activity release Yes Biosphere UNTAMO, SHYD, Pandora, Ecolego Radiological impact MATLAB, ERICA Assessment endpoint Annual activity release Radiation doses Figure 5-5. Computer codes for analysis of radionuclide release scenarios. Codes used for the analysis of repository system scenarios are shown in red and discussed in this Section. Codes used for biosphere assessment are shown in blue and discussed in Section 5.4. For modelling purposes, the radionuclide inventory inside the spent nuclear fuel canisters is assigned to three characteristic locations, namely (i) the fuel matrix, (ii) the grain boundaries and gaps, and (iii) structural materials (zirconium alloys and other metal parts). Following canister failure and contact of the fuel with water, there will be a relatively rapid release to solution of the radionuclide inventory at grain boundaries, in gaps and in corroding metals. This part of the radionuclide inventory is quantified by multiplication of the total inventory by an instant release fraction, or IRF, so-called because the release of this part of the inventory to solution is cautiously modelled as instantaneous. The remainder of the inventory is assumed to be uniformly mixed within the fuel matrix and structural materials, and to be released congruently with their degradation over time. The water inside the canister is groundwater conditioned by the surrounding buffer. Different types of groundwater are considered (e.g. brackish, saline or glacial) according to assumptions inherent in each scenario. In most of the cases, the geochemical conditions and flow distributions are assumed to be time invariant. However, this approach is complemented by calculation cases in which the evolving geochemical conditions and flow distribution (in the variant and disturbance scenarios) are modelled explicitly (albeit in a simplified manner). The explicit modelling of time dependency of flow and geochemical conditions takes account of the adverse possibility that radionuclides may accumulate within the repository system under one set of

161 116 conditions, and then be relatively rapidly transported to the biosphere when conditions change, for example in connection with permafrost episodes. Once released, radionuclides are assumed to dissolve in the water in a canister or to precipitate if their respective solubility limits are reached. Concentrations of all isotopes of the same element, including stable isotopes, are taken into account in determining whether this is the case. The main retention and transport processes in the near field are assumed to be solubility limitation, sorption, diffusion and, in the case of the deposition tunnel and its EDZ, advection. In the majority of calculation cases, radionuclides diffuse from the internal void space of the canister and through the surrounding buffer, retarded by sorption, exiting the deposition hole by three possible paths (Figure 5-6): the F-path, which leads from the canister, through the buffer to a host-rock fracture intersecting the deposition hole; the DZ-path, which leads from the canister to the deposition tunnel EDZ, either directly through the buffer or via a damaged zone, which is assumed to surround the deposition hole, and thence to a host-rock fracture intersecting the EDZ; and the TDZ-path, which leads from the canister, through the buffer to the deposition tunnel backfill, and thence to a host-rock fracture intersecting the deposition tunnel. In cases in which the canister fails due to the presence of a penetrating defect, the diffusive transport resistance of the defect is taken into account. In a few calculation cases, the buffer is assumed to be eroded over time, in which case advection is assumed to be the dominant transport process in the eroded part of the buffer. Nearfield release, retention and transport modelling is performed using the transport module of the GoldSim computer code. The calculated radionuclide release rates via each the three paths listed above provide input to geosphere transport modelling.

162 117 Deposition tunnel Deposition hole EDZ Failure location Canister Water-conducting fractures F-path DZ-path TDZ-path Damaged zone Figure 5-6. Main features and physical dimensions of the near-field model in a vertical section passing through the centre of a deposition hole parallel to the deposition tunnel axis (blue arrows denote the water flow in the deposition tunnel backfill). The main transport process in the geosphere is assumed to be advection along fractures, retarded by matrix diffusion and sorption (Figure 5-7). The transport paths and their properties flow rate, the fraction of the flow path belonging to a specific fracture type and discharge location are based on the groundwater flow modelling supporting the analysis of the radionuclide releases scenarios. The fracture types are defined based on fracture data from the Olkiluoto site and consist of: clay (and possibly sulphide) coated fractures; calcite (and possibly clay and sulphide) coated fractures; slickensided 19 fractures; and other fractures. 19 Slickenside is a polished fault surface formed by frictional wear during sliding, but now used to denote any of several types of lineated fault surfaces.

163 118 These fracture types differ in the characteristics (thickness, porosity and effective diffusion coefficient) assumed for the fracture coatings and adjacent rock matrix layers, and hence in the degree of retention by matrix diffusion and sorption experienced by migrating radionuclides. The definition of the sorption parameters is discussed in Models and Data for the Repository System. The values used in the assessment are selected from the available sorption data for Olkiluoto-specific rocks and groundwaters. The geosphere transport modelling provides, as output, release locations and release rates to the biosphere, i.e. the geo-bio fluxes, which are used as input in the biosphere assessment. Geosphere transport modelling is carried out using either GoldSim (for stochastic simulations and for simplified modelling carried out for the rock shear scenarios RS and RS-DIL) or the MARFA code (for other deterministic calculation cases). Stochastic simulations have been carried as part of the probabilistic sensitivity assessment to study the sensitivity of the model output to variations in the input parameter values. The analysis considered the total release rates from the near field to geosphere and from geosphere to biosphere (summed over the three paths, see above) and applied Monte Carlo methods, graphical methods and variance decomposition methods (see Cormenzana 2013a for details). Figure 5-7. The main retention and transport processes in a water-conducting feature.

164 119 Table 5-4. Main models and modelling tools used for the analysis of radionuclide release scenarios for the repository system. Models and codes Purpose and use Models for the analysis of radionuclide release scenarios for the repository system Models take into account the following processes: radioactive decay, radionuclide release, solubility limitation, sorption, diffusion and advection (and dispersion in the geosphere, see Figure 5-6 and Figure 5-7). The codes applied are summarised below. The activity fluxes provided as output are used to address compliance with the regulatory requirements or are used as input for the biosphere assessment. Main reference: Assessment of Radionuclide Release Scenarios for the Repository System. GoldSim The transport module of GoldSim is used for analysis of the near-field release, retention and transport, also for geosphere retention and transport modelling in stochastic simulations and in the simplified geosphere modelling carried out for the rock shear scenarios RS and RS-DIL. The GoldSim near-field model addresses radionuclide release from a failed canister; and radionuclide transport through the repository near field to the geosphere resulting in the release from the near field to the geosphere The input to near field modelling includes radionuclide inventory and half lives, amount of fuel in the failed canister, fuel dissolution rate, corrosion rates of zirconium alloy and other metal parts, flow properties around the deposition hole containing the failed canister, properties of the canister defect (if any), density, porosity and and elemental solubilities, diffusion coefficients and distribution coefficients in the buffer and backfill dependent on the assumed groundwater composition and flow rates according to the CONNECTFLOW model. The geosphere transport model implemented in GoldSim addresses the migration of radionuclides from each of the entry points (see Figure 5-6) through the geosphere fracture network. Inputs to geosphere modelling include the transport resistance of the geosphere (WL/Q) and the porosity, elemental distribution coefficients and diffusion coefficients in the fracture coatings and rock matrix. Flow-related parameters (e.g. near-field flows and WL/Q) are, in general, based on the results of groundwater flow modelling using CONNECTFLOW (see Table 5-2). The main output from near-field modelling is the radionuclide release rates to the geosphere along the F-, DZ- and TDZ-paths. The main output of geosphere modelling is the spatially-integrated radionuclide release rates to the biosphere (the geo-bio fluxes). MARFA MARFA is used for analysis of radionuclide retention and transport in the geosphere in most of the deterministic calculation cases. The geosphere transport model implemented in MARFA addresses the migration of radionuclides from each of the entry points (see Figure 5-6) through the geosphere fracture network. The variation of the geosphere transport resistance (WL/Q) along the migration paths is read directly from the result of groundwater flow modelling (CONNECTFLOW, see Table 5-2). The release rates from the near field are based on GoldSim calculations. Other inputs are the same as described above for geosphere retention and transport modelling using GoldSim The main output of geosphere modelling is the spatially-integrated radionuclide release rates to the biosphere (the geo-bio fluxes) and also the release locations associated with the F-, DZ- and TDZ-paths. 5.4 Models and data for the biosphere assessment An overview of the biosphere assessment process is given in Figure 5-8. The biosphere description sub-process includes environmental studies and monitoring at the site, compilation of a scientific synthesis of the current state of the surface environment at the site (Biosphere Description) and production of site- and regional-specific data for the safety analysis (Biosphere Data Basis). Formulation of scenarios defines the lines of

165 120 evolution and the calculation cases to be addressed. In the following, the modelling carried out within the four remaining sub-processes, surface environment development, screening analysis, landscape modelling and radiological impact analysis is discussed. A summary of the models and codes used in the biosphere assessment is given in Table 5-5. The radionuclide release and transport models, the main supporting models and information flows between these models are shown in Figure 5-4 and the codes applied to the analysis of radionuclide releases and transport in Figure 5-5. Site characterisation* Biosphere description Environmental studies and monitoring Past & future* development Current state of the surface environment FEPs* and narrative lines of evolution Site & regional data compilation Climatic conditions Formulation of scenarios Credible lines of evolution Scenario drivers * Formulation and classification of scenarios Definition of calculation cases Climatic conditions Geosphere modelling Geosphere modelling Surface environment development Screening analysis Terrain evolution extrapolated from past development Typical succession lines and resource driven land use Conceptual models for elemental circulation and accumulation Projections of the surface environment development Surface and near-surface hydrological modelling Stylised representation of the surface environment Highly cautious transport and dose models* Highly cautious selection of parameter values Screening out radionuclides for further assessment Geosphere modelling Landscape modelling Simplified representation of the surface environment Hydrologically connected biosphere objects Cautious transport models* Time-dependent radionuclide-specific activity distributions Radiological impact analysis Maximum use of local resources Doses to each individual (full dose distributions) Typical absorbed dose rates to plants and animals Annual doses to most exposed and other people * Experience and knowledge from previous iterations of the biosphere assessment important Figure 5-8. The biosphere assessment process, its sub-processes (blue boxes) and key features and products of each subprocess. The green boxes show main inputs from other activities in the TURVA-2012 safety case.

166 Development of surface environment The objective in this sub-process is to project the future development of the surface environment by extrapolating the past regional development, using knowledge of the present conditions in the surface environment and the understanding of relevant FEPs. The set of projections produced in the present assessment is intended to bound the credible trajectories of the future, from a radiation exposure of humans, plants and animals perspective. The future development of the surface environment is assessed using terrain and ecosystems development modelling (TESM, see Terrain and Ecosystem Development Modelling for details) and surface and near-surface hydrological modelling (SHYD, see Surface and Near-surface Hydrological Modelling for details) (see Figure 5-8 and Table 5-5). In the TESM, land-uplift-driven changes and other changes in the surface environment are simulated, until and beyond the time when the potential releases would reach it. The projections use typical succession lines for the development of sea bottom, shoreline, forests, mires, lakes, small water bodies and, rivers. Humans are assuming to use the landscape based on the resources it provides and on needs, typically taking the most suitable (in terms of profit, accessibility, etc.) first resources into use. This gives a link to the exposure pathways that are defined to represent present-day human habits. Natural developments and changes in how humans use the land result in a projection containing distinct biotopes, in which fauna find their habitats. The food web or the structure of the biotic community can then be outlined and representative species and their habits identified. A sub-set of the projections produced by TESM is selected for propagation to landscape modelling. The biosphere objects relevant to these projections are identified and characterised. A biosphere object describes a continuous and reasonably homogeneous segment of the modelled area into which radionuclides may be released. The contamination can take place either by direct release of radionuclides from the geosphere or by horizontal transport of radionuclides within the surface environment during the dose assessment time window. Each biosphere object is characterised by one or more biotopes and a set of object-specific parameters. The biotopes can be divided into two main sub-groups: terrestrial and aquatic biotopes, refined further, e.g. to forest, cropland with varying products, rivers and open sea. The SHYD model is a tool that can be used to study the water balance components at the Olkiluoto site. The model links the unsaturated and saturated soil water in the overburden and groundwater in the bedrock as a continuous system. The fluxes for the biosphere assessment are calculated in two steps: first steady-state recharge/discharge to/from bedrock is computed for each time step; and in the second step vertical and horizontal fluxes are computed for each delineated biosphere object. These fluxes are averages for the specific biosphere objects from the results of the full 3D-model. A new feature of the SHYD modelling compared with previous assessments is that (shallow) wells (both those dug in the overburden and drilled in the bedrock) can be added as sink points in the computational grid.

167 Screening analysis Experience from previous assessments shows that the calculated individual radionuclide-specific releases from the geosphere span several orders of magnitude, both in activity and radiotoxicity, and that, ultimately, the potential radiological impact is dominated by a small number of radionuclides, as commonly identified through cooperation in BIOPROTA 20. The screening models are implemented with a high degree of conservatism to ensure that the calculation results undoubtedly overestimate any potential radiological impacts. The screening analysis applied is in line with a graded approach for safety assessments (IAEA 2009, Requirement 1). The graded approach is implemented by employing a three-tiered modelling approach for radionuclide transport and dose calculations, where the complexity and realism are greater for higher tiers than for lower tiers. Tier 1 and Tier 2 (described below) are generic radionuclide screening analyses requiring a minimum of site-specific data. Tier 3 is based on landscape modelling (Section 5.4.3) and the radiological impact analysis described in Section The aim of the screening analysis is to identify radionuclides that will with high confidence have insignificant radiological impact, and hence can be screened out from analysis with the landscape model. The screening analysis is described in more detail in Biosphere Radionuclide Transport and Dose Assessment Modelling. Tier 1. Radiotoxicity screening analysis In Tier 1 it is assumed that a hypothetical person is exposed, via ingestion, to the total activity released from the geosphere during the whole dose assessment time window. Hence, this hypothetical person is exposed to the entire radiotoxicity of the released activity, which is unarguably very cautious. Tier 2. Generic biosphere model screening analysis The model (Section of Biosphere Assessment) used in Tier 2 analysis has a higher degree of realism than the Tier 1 model, but is still sufficiently cautious for screening purposes. It contains a set of mainly generic ecosystem-specific sub-models, similar to the screening models proposed in IAEA (2001) that may be used to determine through a simplified but cautious assessment whether a radionuclide can be neglected from further consideration. The sub-models in Tier 2 comprises a water well, a lake, a forest, an irrigated cropland and an irrigated pasture land. Exposure pathways considered are ingestion of radionuclides in water, milk, crops, livestock meat, game, mushrooms and berries, inhalation of radionuclides and external exposure from radionuclides in the ground. The screening decision for a specific radionuclide is based on the sub-model resulting in the highest calculated dose Landscape modelling The landscape model is a state-of-the-art, time-dependent and site-specific radionuclide transport model that takes the properties of the dynamic site into account. 20 International collaboration forum which seeks to address key uncertainties in the assessment of radiation doses in the long term arising from releases of radionuclides as a result of radioactive waste management practices (

168 123 The landscape model is constructed from the biosphere objects delineated in the surface environment development sub-process (Section 5.4.1). Defining the initial state for the landscape model and how it develops with time is the landscape set-up activity, which is the task interfacing the surface environment development and the landscape modelling (see Figure 5-4). The connections between the biosphere objects are derived from terrain projections for the period from the present (initial state) to the end of the dose assessment time window. These connections determine how the biosphere objects are hydrologically connected, i.e. if one object is upstream or downstream of another. Each ecosystem type in each biosphere object is associated with a deterministic radionuclide transport compartment model (see Figure 5-9, left). An important feature of the landscape model is that it represents ecosystem succession, by allowing transitions over time from one ecosystem type to another. The underlying radionuclide transport models within biosphere objects must be consistent. Thus, it is important to ensure: 1) that the water fluxes between biosphere objects can be handled in a continuous manner, and 2) that the activity content of a specific compartment in a specific biosphere object can be transferred to a corresponding compartment when an ecosystem type develops into another type. Within the landscape modelling concept, the biosphere objects form aggregates which are called super-objects (Figure 5-9, right). The super-objects are within landscape zones delineated so that their area does not change with the landscape development, i.e. the transport of radionuclides due to change of the ecosystem types is confined within the given zone. This facilitates the maintenance of mass balance in the model and prevents transitional effects occurring as landscape changes. Terrestrial blocks Mire Upland forest Cropland irrigated Cropland Non-irrigated Aquatic blocks Reed bed Open water Figure 5-9. Left: the common structure for modelling any ecosystem type in a biosphere object. Right: The general ecosystem type structure of super-objects. Out

169 124 The main output of the landscape modelling sub-process is time-dependent radionuclide-specific spatial activity distributions in all biosphere objects for the analysed calculation cases. These are the key input to the radiological impact analysis Radiological impact analysis In the radiological impact analysis, the spatially distributed, time-dependent radionuclide-specific activity concentrations in environmental media, produced by landscape modelling, are used to calculate the potential radiological impact, in terms of annual doses to humans and absorbed dose rates to plants and animals. The calculated annual doses and absorbed dose rates are assessed against the radiation protection criteria defined by the STUK (see Table 1-1). Assessing doses to humans The regulations require assessments for two exposed groups: annual dose to the most exposed people and average annual doses to other people. The dose assessment method is based on the deterministic approach developed and applied in Broed et al. (2007) and further refined in the interim safety case (Hjerpe et al. 2010). This concept has been further developed for this assessment, to include consideration of shallow drinking water wells and consumption of farm animal products. It is based on the guidance given by the ICRP (2006) on assessing the dose to the representative person (see Biosphere Radionuclide Transport and Dose Assessment Modelling for details). The key information needed for the dose assessments is radionuclide concentrations in environmental media, such as soils, sediments, surface waters, well water and air. This information is derived from landscape modelling. Production rates of individual food products are also needed, to derive the total productivity, and concentration ratios of radionuclides from environmental media to foodstuffs. The exposure characteristics are mainly based on site-specific conditions, regional land use, and present-day behaviour of the regional population. In long-term safety assessments of geological disposal facilities it is common that a few exposure pathways dominate the calculated doses. The following pathways are considered in the modelling: ingestion of foodstuffs from aquatic, terrestrial and agricultural ecosystems, ingestion of drinking water from freshwater surface waters and wells, inhalation of airborne contaminants and external radiation from radionuclides in the ground. The dose calculations for foodstuffs are based on values of food energy intake, for individuals making the greatest reasonable use of the most contaminated local resources. The dose coefficients for ingestion and inhalation are values recommended for adults (ICRP 1996). Calculations of external radiation doses from contaminated soil and sediments use dose coefficients for radionuclides uniformly distributed to an infinite depth (based on Table III.7 in EPA 1993, extracted using the software Radiological Toolbox 21 ). 21 U.S. Nuclear Regulatory Commission Radiological Toolbox, (version 2.0.0, August 2006) (

170 125 The present assessment calculates landscape doses, which are pathway-, radionuclideand biosphere object-specific annual doses to a person. These are then combined to obtain an annual landscape dose, by summing the dose maxima from each pathway. Then the annual landscape dose is calculated for each exposed person to yield a distribution of doses called the dose distribution. It is assumed that the exposed population is no larger than the size of the present-day population of Eurajoki municipally (about 6000 persons); unless the calculations result in that the production of contaminated foodstuffs may support an even larger population, then the population size is increased. Furthermore, it is assumed that the exposed population obtains all their drinking water from contaminated water sources, consumes all contaminated foodstuffs from the whole modelled area and inhabits all suitable residential areas. The dose distribution is calculated for each generation using the contaminated area during the assessment time window. The dose distributions, derived for each generation using the contaminated area during the dose assessment time window, are then divided into two groups: the most exposed group and other people. The most exposed group is identified as a sub-group in the exposed population that receives the highest doses; a size of 20 persons is considered appropriate for this group. The group of other exposed people is then taken as the whole exposed population, excluding the most exposed group. The average doses in these two groups are then calculated and provide the annual doses to representative persons for the most exposed people and for other people, as required by the STUK Guide YVL D.5. Assessing doses to plants and animals The first step in assessing dose to plant and animals is to identify representative species. As new land areas will form over time as a result of continued post-glacial land uplift, lake and inland mire biotopes not currently present at Olkiluoto will arise. Species representative of such biotopes have been identified through Posiva s survey and monitoring program of reference lakes and mires (see Biosphere Description). The process by which representative species have been selected for assessment and the means by which they have been parameterised to allow dose calculations to be performed are described in the following. The diversity of plants and animals in terrestrial and aquatic ecosystems is such that it is not possible to consider all species of plant and animal explicitly within an assessment (Ulanovsky & Pröhl 2008). Simplification is required to allow dose implications from the long-term releases from the proposed Olkiluoto repository to be evaluated. Such simplifications are implicit within available assessment approaches and tools, including the ERICA (ERICA 2007) and the ICRP Reference Animals and Plants (ICRP 2008) approaches. A sensitivity and knowledge quality assessment, as reported in Smith et al. (2010), identified conceptual uncertainties associated with the application of the ERICA assessment method to post-closure assessments. Based on the information in Smith et al (2010), the following criteria were considered to be relevant to the selection of representative biota for the Olkiluoto assessment: Species that are strongly identified with the reference area (or that can be reasonably predicted to migrate into new biotopes as they are formed through land emergence

171 126 and natural succession identified from the monitoring programme for reference biotopes); Species considered having a greater exposure potential, e.g., organisms with soildwelling habits, for which external exposure may be maximised; Organisms occupying likely groundwater discharge areas, particularly water-body margins and wetlands (mires), etc. Representation of the different trophic levels as depicted in biotope-specific food webs; Species that are socioeconomically important and therefore of particular public interest; and Species for which there is reasonable knowledge relating to behaviour within ecosystems and for which data on radionuclide transport were considered likely to be available either from site-studies or the broader international literature (or for which analogue data could reasonably be applied). Each thus selected organism is represented by an ellipsoid of dimension and mass commensurate with that organism. Ellipsoids are then used as a means of calculating dose conversion coefficients (DCC) that enable absorbed dose rate to be calculated. Two sets of DCC s are applied: DCC ext relates the activity concentration of a radionuclide in environmental media (soil, sediment, water or air) to the external absorbed dose rate (µgy/h) received by the organism in relation to its occupancy habits in the environmental media; DCC int then provides the mechanism by which internal absorbed dose rate (µgy/h) can be calculated in relation to radionuclides within the body of the organism. The value of the DCC, both external and internal, is dependent upon the size and geometry of the organism and its position relative to environmental media. DCC s for each representative organism were calculated using the add reference organism functionality within the ERICA assessment tool (see details in Dose Assessment to Plants and Animals). The approach to deriving typical dose rates for representative species is broadly commensurate with that applied in deriving average annual dose to people. It uses the DCC s discussed above in conjunction with the same radionuclide concentrations in environmental media as used in calculating doses to humans. Typical dose rates are calculated by deriving an area-weighted dose rate for each representative species exemplar throughout the contaminated biosphere objects in the modelled area. The area weighted dose rate is computed for each 50 year assessment time-step (i.e. biotope progression throughout the assessment timeframe is taken into account). For species transient between different biotopes, the same approach is taken to calculate the area weighted dose rate per biotope. The relative biotope occupancy is then applied to calculate the dose rate according to time spent between the different biotopes.

172 127 Table 5-5. Main models and codes used for the biosphere assessment in safety case TURVA Models and codes Purpose and use Models and codes for the development of the surface environment Projecting the development of the surface environment is implemented in two modelling activities: the terrain and ecosystems development modelling (TESM) and the surface and near-surface hydrological modelling (SHYD). In the TESM, land-uplift-driven changes and other changes in the surface environment are simulated, until and beyond the time when the potential releases would reach it (see UNTAMO below). The TESM and SHYD modelling are documented in detail in Terrain and Ecosystem Development Modelling and Surface and Near-Surface Hydrological Modelling, respectively. UNTAMO UNTAMO is a multi-module geographical information system (GIS) toolbox that has been developed for projecting the development of the terrain and ecosystem during the dose assessment time window. Each module in the toolbox addresses a certain aspect of the surface environment development: land uplift and delineation of the sea area, surface water bodies, erosion and deposition, terrestrial and aquatic vegetation and land use. The UNTAMO toolbox is used together with the SHYD model. The terrain and ecosystems are projected into future with UNTAMO and delivered as input data to the SHYD model to simulate groundwater flow and water table characteristics in detail; on the other hand, UNTAMO uses a simplified groundwater table model derived from an earlier iteration with SHYD to identify areas of shallow and deep groundwater table. SHYD The SHYD model estimates the movement and storage of water in the radionuclide transport models of the surface environment, including horizontal and vertical water fluxes in the overburden and at the ground surface. The model uses a 3-D grid with various types of spatial and temporal simplifications (conceptualisations) linking the unsaturated and saturated soil water in the overburden and groundwater in the bedrock as a continuous pressure system. Key data used are documented in Biosphere Data Basis and summarised in Table 5-6 of Biosphere Assessment. Models and codes for the screening analysis In the screening analysis, a two-tiered (see text) approach is employed, in order to to identify radionuclides that are highly confidently expected to have insignificant radiological impact, and hence can be screened out from further analysis with the complex landscape model. The screening models are implemented with a high degree of conservatism to ensure that the calculation results undoubtedly overestimate any potential radiological impacts. Ecolego The screening analyses were carried out using the software package Ecolego ( which is a simulation software tool used for creating dynamic models and performing deterministic and probabilistic simulations. The screening models are documented in detail in Biosphere Radionuclide Transport and Dose Assessment. Models and codes for the landscape modelling The landscape model is a state-of-the-art, time-dependent and site-specific radionuclide transport (RNT) model that takes the properties of the dynamic site into account. The main outputs of the landscape modelling sub-process are time-dependent radionuclide-specific spatial activity distributions in all biosphere objects for the analysed calculation cases. These are the key input to the next sub-process, the radiological impact analysis. The landscape model is documented in detail in Biosphere Radionuclide Transport and Dose Assessment. Pandora Pandora is a tool developed by Facilia AB for Posiva and SKB and used by Posiva for radionuclide transport modelling in the surface environment. Pandora is based on the Matlab/Simulink environment ( Pandora was developed to simplify development of models resulting in large systems of differential equations where decay of radionuclides is included in the model and to enhance the graphical user interface to be more suitable for radioecological modelling. Models and codes for radiological impact assessment

173 128 Models and codes Purpose and use The aim of the models for radiological impact assessment is to produce estimates for appropriate radiation exposure quantities. Spatially distributed, time-dependent radionuclide-specific activity concentrations in environmental media, produced by the landscape modelling, are used to assess the potential radiological impact, in terms of annual doses to humans and absorbed dose rates to plants and animals, based on present-day radiation protection criteria. The model to calculate annual doses to humans is documented in detail in Biosphere Radionuclide Transport and Dose Assessment and the model to calculate absorbed dose rates to plants and animals is documented in detail in Dose Assessment for Plants and Animals. MATLAB The dose calculations are performed using models implemented in Matlab that utilise the environmental activity concentrations resulting from the landscape modelling. Annual doses to humans are derived for each exposed person individually taking all relevant exposure pathways into account, such as: food ingestion from crops, animal products, fish, game, mushrooms and berries, ingestion of drinking water from wells and freshwater surface bodies, inhalation of breathing air and external exposure from the ground. Absorbed doses rates to plants and animals are derived for each representative species exemplar occupying individual biosphere objects throughout the modelled area. ERICA The ERICA Tool has a structure based upon the ERICA tiered approach to assessing the radiological risk to other biota. The tool is here used to derive dose conversion coefficients via external and internal radiation to reference organisms characterised by default attributes relating to radioecology and dosimetry. The key attributes are equilibrium concentration ratios, occupancy factors, and ellipsoidal geometries.

174 129 6 PERFORMANCE ASSESSMENT OF THE REPOSITORY SYSTEM This chapter summarises the performance of the repository system and the fulfilment of performance targets and target properties for the engineered barriers and host rock as identified in Table 2-2 and Table 2-3. The performance assessment takes account of the expected thermal, hydraulic, mechanical and chemical (THMC) evolution of the repository system, and uncertainties in the expected lines of evolution. Unlikely lines of evolution, including the possibility of disruptive events, are also identified. Account is taken of the natural evolution of the environment, chiefly driven by climatic evolution, which imposes external loads on the repository system, and also internal loads, chiefly from the effects of excavation and emplacement of the spent nuclear fuel and the engineered barriers. The assessment considers the different evolutionary processes that potentially can affect the performance targets and target properties for the three time windows; the period of excavation and operation up to closure, the post-closure period during the next 10,000 years and beyond 10,000 years over repeated glacial cycles up to million years. The fulfilment of performance targets and target properties in each time window is assessed considering time-dependent and space-dependent loads on the engineered barriers and host rock. Quantitative assessments are made whenever possible, e.g. to calculate safety margins and demonstrate the robustness of the design. Uncertainties are highlighted, conditions that could lead to deviations from performance targets and target properties are identified, and the likelihood and effects of the deviations estimated. In particular, conditions and events (incidental deviations) that could lead to the release of radionuclides are identified. These deviations from the desired initial state or expected evolution are propagated to Formulation of Radionuclide Release Scenarios, which defines the scenarios and the calculation cases for both the repository system and the surface environment. 6.1 Excavation and operation up to closure of the disposal facility The excavation and operational period of the underground disposal facility may potentially affect the long-term repository performance, since the changes in the thermal, mechanical, hydrological and chemical conditions induced by the excavation and operational activities may affect the engineered barriers and the host rock in the longer term. The duration of this period in various parts of the repository can be assumed to be from several tens up to about one hundred years, depending on the progress of the excavation and operational activities and the total number of canisters to be disposed Repository system evolution and performance Hydraulic and geochemical evolution of geosphere Groundwater flow at Olkiluoto takes place mainly through a network of fractures and deformation zones, within which channelling of the flow is likely. Thus, there will be significant local variation of the flow conditions and possibly also of salinity and groundwater composition near the deposition holes.

175 130 During the operational period, the flow rates are approximately two orders of magnitude higher than before the repository construction (see Figure 6-1). Following installation of tunnel and shaft backfill and seals (i.e. after closure), modelled flow rates return to near pre-excavation values. During the operational period, the open tunnels draw water from all directions so that the flow below the repository is upwards compared with the mainly downward flow in the natural state. The upward flow is somewhat strengthened by the heat produced by the spent nuclear fuel. The correlation between the inflow and post-closure flow rate is not necessarily uniform and may vary spatially. Thus it is possible that even if all the deposition holes with inflow over the maximum inflow criterion to a deposition hole defined by RSC, 0.1 L/min, are discarded, it cannot be excluded that a few deposition holes will be associated with a higher post-closure flow rate or a lower transport resistance than the target values (see Table 2-3). Figure 6-1. Flow rate into and out of the reference volume containing the repository (Löfman & Karvonen 2012; 2009SH refers to a model variant with layout for 5500 tu and the hydrogeological model described in the previous site description (Posiva 2009b) with semi-homogeneous (SH) hydraulic properties of the hydrogeological zones (HZ) and sparsely fractured rock (SFR) i.e. for most of the HZs homogeneous properties are used, whereas the SFR is divided into the depth intervals, in which either depth-dependent or homogeneous values are used. 2011SH refers to a model with layout for 9000 tu and the hydrogeological model according to Site Description with the hydraulic properties of HZs and SFR described as in 2009SH, and 2011HE refers to a model with layout for 9000 tu and the hydrogeological model according to Site Description with the heterogeneous (HE) hydraulic properties of HZs and SFR (Performance Assessment, Chapter 5).

176 131 Excavation will cause a damaged zone (EDZ) to form, especially below the tunnel floors, although the damage is unlikely to form a continuous hydraulic pathway along the tunnel length (Site Description, Section ). In addition, excavation, and later the heat produced by the spent nuclear fuel, may cause spalling or other types of stressinduced damage around the excavated rooms, deposition tunnels and deposition holes. There are minor uncertainties related to the extent and properties of the EDZ and the stress-induced rock damage. These uncertainties are taken into account in the groundwater flow modelling. The potential presence of an EDZ affects inflows below 1 ml/min and means that all deposition holes have some, although in many cases very limited inflow compared with 40 % of deposition holes with no inflow when there is no EDZ present. During the operational period, the average salinity around the repository remains similar to the pre-construction phase and changes are rather moderate (Figure 6-2). However, increased groundwater flow into the repository volume may lead to mixing of water and to either more dilute or more saline conditions locally. However, even under pessimistic assumptions, the maximum salinities in the reference volume are expected to remain below 70 g/l and salinities over 35 g/l are not expected at repository depth. Most of the modelling results suggest that the lowest salinities during the operational period will be at least a few grams per litre in most parts of the repository. However, the possibility of salinities close to 0.3 to 0.4 g/l (which corresponds roughly to the minimum acceptable total charge equivalent of cations of 4 mm) cannot totally be excluded. Such low salinities are obtained only under pessimistic assumptions with infiltrating water of zero salinity and neglecting the geochemical and hydrochemical reactions in the overburden and along the recharge paths. The lowest and highest salinities discussed above are related to the main hydrogeological zones in the ONKALO and would not necessarily occur in the repository panels themselves, which avoid hydrogeologically active zones. Moreover, the disturbed conditions of either low or extremely low salinities are likely to last a limited time in the order of a few tens of years, and thus the impact on the performance of the buffer and backfill is limited. In summary, in spite of the rather large variations in the flow conditions during the excavation and operational period, the groundwater composition with respect to salinity, chloride content and total charge equivalent of cations will remain within the target range meaning that the buffer and backfill functions are fully preserved, except at a few canister positions. Organics, pyrite and other sulphides in the overburden are able to consume oxygen in infiltrating waters by microbially mediated reactions. Thus, oxygen is not expected to penetrate more than a few tens of metres along fractures and is very unlikely to reach repository depth. The ph in the infiltrating water is neutralised by calcite. The ph in the natural groundwater at the repository depth is expected to be in the range of 7 to 9 and is thus well within the range defined by the target properties.

177 132 Figure 6-2. Salinity (TDS) evolution during the excavation and operation period, and after closure until 50,000 years after the start of construction of the ONKALO; the maximum, minimum and average salinity in the reference volume for model variants 2009SH, 2011SH and 2011HE, see Figure 6-1 for explanations (from Performance Assessment, Chapter 5). The Olkiluoto groundwater has a naturally low colloid content. Colloids may be formed by cement or silica sol degradation. However, as the groundwater at the repository depth generally has such a high ionic strength (salinity), such colloid formation is expected to be limited. The sulphide concentration for the main water types in the natural (undisturbed) state is well below 1 mg/l due to the control of sulphide concentration by iron in the groundwater forming iron sulphides, which have low solubility. It has been observed that site characterisation activities and ONKALO construction have caused artificially disturbed transient conditions due to mixing of different groundwater types and anomalous sulphide levels have been measured (max. 12 mg/l) at a depth of around 300 m. The high concentrations of sulphide are probably due to a delay in the availability of iron; however, sulphide concentrations are still evidently controlled by iron sulphide phases in the longer term. According to monitoring results, sulphide concentrations decrease from the anomalously high values once the groundwater conditions stabilise. Although the groundwater data clearly indicate sulphide values well below 1 mg/l, a pessimistic upper value of 3 mg/l is used in Performance Assessment, for the whole assessment period, which accounts for the possibility of solubility control by the more soluble amorphous iron sulphide in combination with kinetically-constrained availability of iron and the uncertainties related to microbial activity.

178 133 Thermal evolution The calculated temperature evolution at the canister surface and at the deposition hole wall is illustrated in Figure 6-3 assuming either unsaturated or saturated buffer. The maximum temperature at the canister surface is 95 C at initial state, when there is a dry air gap between the canister and the buffer and the buffer is unsaturated, and it is achieved within about 20 years after canister emplacement. In the case of a saturated buffer, the temperature at the canister surface is at most about 75 C. The maximum temperature in rock at the deposition hole wall is reached within about 40 years and is about 65 C. The results shown are for a canister located in the central part of the repository, illustrating the maximum temperatures that will be reached. Thus, the maximum temperatures after emplacement and during the operational period will remain within the acceptable ranges. Figure 6-3. Canister surface temperature estimates in the repository (central area) as a function of time since emplacement assuming the two extreme saturation degrees for the bentonite buffer, either unsaturated (buffer in initial condition) or saturated buffer. The temperature evolution of rock at deposition hole wall (buffer-rock interface), which does not depend on the degree of buffer saturation, is also shown. OL3 canister, average burnup 50 MWd/kgU, separation between deposition holes and deposition tunnels of 10.5 m and 25 m respectively, buffer conductivity is 1.0 W/m/K in the initial condition and 1.3 W/m/K in saturated condition. In the initial condition, there is a 10 mm air gap between the canister and the buffer, in the saturated condition the gap is closed. The outer 50 mm gap between buffer and rock is assumed to be filled with bentonite pellets that have conductivity of 0.2 W/m/K in the initial condition and 0.6 W/m/K when saturated. Based on the results of Ikonen & Raiko (2013).

179 134 Rock mechanical evolution Excavation and thermal load caused by the decay heat from the spent nuclear fuel will cause some local damage, e.g. spalling, reactivation of fractures, within the near-field rock. Although rock damage is not directly considered in the target properties, rock damage around the excavated rooms, deposition tunnels and deposition holes may have an impact on the hydraulic properties of the rock (as discussed above). Excavation will cause a damaged zone (EDZ) to form, especially below the tunnel floors, although the damage will probably not be continuous. Due to the uncertainties related to the continuity and hydraulic properties of the EDZ and its impact on the groundwater flow, variations in its characteristics have been studied using alternative assumptions as to its properties in groundwater flow modelling. Although there is a good understanding of the processes affecting the mechanical state of the rock in general, there are still uncertainties related to the elastic and rock strength parameters of the rock at Olkiluoto and especially concerning the in-situ stress state, which need further study. The heterogeneity of the rock leads to spatial variations in rock properties, which complicate the assessment of rock damage. To cope with the uncertainties as to the existence of rock damage around deposition holes, two cases have been considered in the groundwater flow modelling: i) a hydraulically significant rock damage zone around the deposition hole and ii) no hydraulically significant rock damage around the deposition hole. The hydraulic conductivity of the damage zone has also been varied. Reactivation of fractures can change their hydraulic properties, but the changes in the hydraulic properties are minor, especially when compared with changes caused by the formation of the EDZ around the tunnels and rock damage around the deposition holes. Mechanical and hydraulic evolution of buffer and backfill Before full saturation, some buffer and backfill material may be lost through piping and erosion. Based on the calculated inflow to deposition holes, roughly one third of the deposition holes are such that some limited buffer mass loss is expected due to piping and erosion. In a base case, considering the maximum inflow criterion of 0.1 L/min in a deposition hole, the estimated mass loss is at most 185 kg (the total buffer mass in one deposition hole is 20,300 24,300 kg). There are variant cases with larger losses, but the average buffer density remains such that no significant changes are expected in the hydraulic conductivity or the swelling pressure of the buffer and the necessary low hydraulic conductivity and sufficient swelling pressure will be achieved as the buffer saturates. It is estimated that at most 13,000 kg of the backfill could be locally lost by piping and erosion, but the eroded material would be redistributed within the deposition tunnel. This is rather small compared with the total mass of backfill material in the tunnel (more than 8000 tonnes in a 300 m long deposition tunnel). The effect on the backfill performance depends on how the mass loss is distributed in the backfill. For example, if all of the 13,000 kg were lost from a tunnel section of 1 m, the mass loss would have a significant effect on the backfill density at this location. Such an event could perhaps be possible in the vicinity of a fracture with a high enough inflow to transport all this mass further down in the tunnel. It is also possible, after erosive mass redistribution is completed, that homogenisation over time may mitigate the localised mass loss to some

180 135 extent. In any case such erosion would not be detrimental to the performance of the EBS system, since no deposition hole would be located near such a fracture according to the RSC. In conclusion, the buffer and backfill will maintain properties consistent with their performance targets even considering the process of piping and erosion. The remaining uncertainties of a situation with a significant loss of buffer in a deposition hole are considered in the formulation of release scenarios. Geochemical evolution of buffer and backfill During saturation some limited dissolution and precipitation of salts in the buffer may take place due to heat transfer from the canister. However, this is expected to have a limited impact on the buffer. Cementitious leachates from grouting of fractures, from grout used to stabilise rock bolts and from the plug in the deposition tunnel may locally affect the backfill during saturation due to degradation and leaching of cementitious materials. However, no cement is in direct contact with the buffer and thus the impact on the buffer is expected to be of little significance. For both non-saturated and saturated conditions, the consumption of initially present oxygen in the backfill and buffer will be relatively rapid (in the order of a few days to a few years), due to its reaction with pyrite and other accessory minerals. Oxygen is also partly consumed by the canister. Thus, anoxic reducing conditions will be quickly established around the emplaced canisters and throughout the buffer and backfill. Similarly to the natural groundwater colloids and cementitious colloids, the introduced colloids through degradation of buffer and backfill materials are expected to be scarce in the high ionic strength groundwaters during the operational phase and the postclosure period up until much later when the infiltration of dilute meltwater after an icesheet retreat has to be considered (see Section 6.3). Mechanical, hydraulic and geochemical evolution of closure There is a relatively short time (30 50 years) between the start of the emplacement of the first closure components (by 2070) and the finalisation of the closure (by ) and thus there should not be major uncertainties in their behaviour during the operational period, provided that appropriate quality assurance measures are followed. Colloids generated by the degradation of closure materials (or other introduced materials) will be unstable in the high ionic strength groundwaters at repository depth, and hence of negligible impact. Canister corrosion At emplacement the canisters will be covered by a thin layer of corrosion products. The presence of a few defective canisters in the repository cannot be ruled out at this stage of technical development. The maximum corrosion depth from the atmospheric and initially entrapped oxygen is expected to be less than 0.5 mm, and will thus have a negligible impact on the minimum

181 136 thickness of the copper canister wall. Also, during the early aerobic phase in the evolution of the repository environment, the period of susceptibility to stress corrosion cracking is short and it is highly unlikely that the required conditions (redox potential, porewater salinity, interfacial ph and SCC-inducing species) will be present simultaneously. Residual stresses on the surface of the canister (at the weld location) are currently being quantified. Mechanical impacts on the canister The impact from potential canister handling accidents is not a concern in the long-term safety case, since if such accidents happen, the canister will be returned to the encapsulation plant for examination and assessment. If necessary, the canister will be opened and unloaded, and the fuel re-encapsulated in a new canister. To control the canister surface condition, a final control point for surface damage will be established just before the canister is lowered into the deposition hole. Subcriticality Criticality safety analyses are performed for transportation and encapsulation purposes to ensure, with a high margin of safety, that the canister will be in a sub-critical state at the time of emplacement. Criticality during the early evolution of the repository is not possible because the spent nuclear fuel will remain in the same geometrical configuration as in the initial state and no water (a neutron moderator) is present in the canister. Criticality safety analyses show that, even if it is cautiously assumed that the canister is filled with water, the spent nuclear fuel is expected to remain in sub-critical condition (for OL3 canister, the use of burn-up credit is necessary). There are currently no qualified methods for the use of burnup credit for geological disposal purposes but work is ongoing at international level to address this issue Fulfilment of performance targets and target properties As discussed above, for the expected evolution of conditions during the period of operation and closure, the properties of the EBS and host rock will conform to the performance targets and target properties at the end of the operational period with a high degree of confidence. There are some possibilities for incidental deviations, although none of these are expected. Even if all the deposition holes with inflow over the limit of 0.1 L/min are rejected, it is not possible to exclude the possibility that a few deposition holes might experience a higher post-closure flow rate or a lower transport resistance than the target values. For a few canister positions, the groundwater composition with respect to salinity, chloride content and total charge concentration of the cations may, for a short time, be outside the target ranges. Any such deviations are expected to disappear shortly after closure of the disposal facility and will have minimal effects on the backfill, buffer or canister. It is expected that all canisters leaving the encapsulation plant will be intact, but it cannot be ruled out that a few canisters will have an initial penetrating defect that escapes detection. As discussed in Section 3.5, at present, the probability of

182 137 detection can only be based on expert judgement, taking into account of the results from both the non-destructive testing and destructive testing of the weld. With more data becoming available in the future, it is likely that it will be possible to demonstrate that the probability of emplacing more than one canister with an initial undetected penetrating defect is less than one per cent. Of these possible incidental deviations, only the last gives any possibility for release of radionuclides from the EBS during the first several hundreds of thousands of years. Therefore, the case of one or a few canisters with an initial penetrating defect is carried forward to the formulation of release scenarios and assessment of radionuclide releases. 6.2 Post-closure evolution during the next 10,000 years The climate during the next 10,000 years is expected to remain essentially as today, i.e. a temperate climate associated with a boreal ecosystem. Crustal uplift will continue in southern Finland, outpacing global eustatic sea-level rise. Thus, locally, relative sea level will fall, and as a consequence, hydraulic gradients will increase during the first few thousands of years after repository closure. At 1000 to 2000 years after present, the shoreline will have retreated far enough that further changes will not affect the hydraulic gradient nor alter the flow rates in the repository volume. Groundwater flow and groundwater chemistry will recover from the disturbances caused by the excavation, as discussed in Section 6.1, to return to conditions similar to those of the present day. The groundwater flow is governed by the hydraulic gradients caused by the topography and salinity field. The main impact on the groundwater composition will be due to continued infiltration of meteoric waters. The main processes ongoing in the repository during this stage will be water uptake, saturation, swelling and homogenisation of the swelling clays in the buffer, backfill and closure and the gradual decline of the residual heat in the spent nuclear fuel Repository system evolution and performance Hydraulic and geochemical evolution of the geosphere After closure of the disposal facility, the site will recover from the disturbances caused by repository construction, operation and closure. Flow rates will reduce as the drawdown caused by the repository decreases and saturation of buffer and backfill occurs. For the first hundreds of years after closure the heat produced by the spent nuclear fuel increases the flow rates by a factor of 2 to 3 compared with the natural state and enhances temporarily and locally upward flows. Generally the heat trends to result in an upward driving force for the water, but when combined with the stronger natural downward forces the flow remains still mainly directed downwards. The heat production declines to very low levels after the first few thousands of years. Beyond that time, the main factors affecting the hydrogeological and hydrogeochemical evolution of the site will be the continued crustal uplift and the infiltration of meteoric waters. After closure, the flow rates will reduce significantly due to reduced hydraulic gradients. However, in deeper parts of the rock, including the rock around the repository, the flow rates will remain somewhat higher than in the natural state before

183 138 construction, because the slow recovery of the salinity field will affect the flow field for hundreds of years. Discrete fracture network modelling carried out by Hartley et al. (2013b) provides detailed information about the migration paths and flow around the deposition holes. This information is used both to assess whether the target properties are met as well as providing an input to the radionuclide release, retention and transport analysis. The modelling studies quantify the impacts of the tunnel EDZ and the rock damage around the deposition holes as key factors determining local flow rates and other flow-related transport parameters. Figure 6-4 shows the variation of the cumulative distribution of the flow rate (CDF) per unit width in the release location and flow-related transport resistance for the three release path types (for release paths see Figure 5-6), based on discrete fracture network modelling with particle tracking, with exit from the deposition hole to: a host-rock fracture intersecting the deposition hole (F-path), the excavation damaged zone (EDZ) below the tunnel floor DZ-path) or the tunnel backfill above the deposition hole (TDZ-path). Different assumptions on the continuity of the EDZ and on the presence of rock damage around deposition holes are considered, as well as the case of an open crown space in the deposition tunnel. The modelling of the impacts of the excavation damaged zone and the potential rock damage around the deposition holes on the groundwater flow shows that the connectivity of fractures and flow rate around deposition holes is indeed increased, but the effect of the increased connectivity is limited to the deposition holes that are not intersected by flowing fractures at all or are intersected by fractures with low flow rates per unit width (less than 10-4 m 3 /(m a)) and high transport resistance (higher than 500,000 years/m). The effects on the natural fractures are limited, however, and flow rates in natural fractures and the transport resistances in the vicinity of the deposition holes are consistent with target properties except for a few deposition holes. The salinity evolution has been estimated based on the modelling by Löfman & Karvonen (2012) and Trinchero et al. (2013). The salinity evolution is shown in Figure 6-6 based on the two model variants out of the three considered in the modelling by Löfman & Karvonen (2012). Depending on the model variant, the modelling period covered either 10,000 years or 50,000 years, which is the assumed duration of the temperate period before the next permafrost period. All the three model variants resulted in a more or less similar behaviour of the salinity field. Following the reduced flow in the repository volume and the recovery of the downward flow direction, the salinity field recovers from the disturbance caused by the repository excavation. The recovery of the salinity field is slower than that of the flow field. The changes in salinity are faster within and close to the hydrogeological zones compared with sparsely fractured rock, as can be seen from Figure 6-6. In the model variant with heterogeneous properties of the sparsely fractured rock, there is more variation in

184 139 fraction Base Case No EDZ Continuous EDZ No Spalling Crown Spacing log 10 (U)[m 3 /m,a] fraction Base Case Continuous EDZ No Spalling Crown Spacing log 10 (U)[m 3 /m,a] fraction Base Case No EDZ Continuous EDZ No Spalling Crown Spacing log 10 (U)[m 3 /m,a] Figure 6-4. CDF plots of initial flow-rate, U, for particles reaching the model top boundary, with RSC inflow screening applied. Results are shown for QF (top), QDZ (middle) and QTDZ (bottom) release paths for a number of variants with different assumptions as to the EDZ and rock damage around the deposition hole as well as a case assuming that the top part of the backfill has high conductivity (presence of a crown space). In the base case, a discontinuous EDZ and rock damage around the deposition hole are assumed. (Note: for the QDZ release path there is no CDF for the No EDZ case.)

185 140 fraction Base Case No EDZ 0.8 Continuous EDZ 0.7 No Spalling 0.6 Crown Spacing log 10 (F r )[a/m] fraction Base Case 0.3 Continuous EDZ 0.2 No Spalling 0.1 Crown Spacing log 10 (F r )[a/m] fraction Base Case 0.3 No EDZ 0.2 Continuous EDZ No Spalling 0.1 Crown Spacing log 10 (F r )[a/m] Figure 6-5. CDF plots of flow-related transport resistance, Fr (=2WL/Q), for particles reaching the model top boundary, with RSC inflow screening applied. Results are shown for QF (top), QDZ (middle) and QTDZ (bottom) release paths for a number of variants with different assumptions as to the EDZ and rock damage around the deposition hole as well as a case assuming that the top part of the backfill has high conductivity (presence of a crown space). In the base case a discontinuous EDZ and rock damage around the deposition hole are assumed. (Note: for the QDZ release path there is no CDF for the No EDZ case.)

186 141 Model variant 1 (2009SH) Layout for 5500 tu and the hydrogeological model 2009 with semi-homogeneous (SH) hydraulic properties of the hydrogeological zones (HZ) and sparsely fractured rock (SFR) t = 1000 years Model variant 3 (2011HE) Layout for 9000 tu and the hydrogeological model 2011 with the heterogeneous (HE) hydraulic properties of of the hydrogeological zones (HZ) and sparsely fractured rock (SFR) t = 10,000 years t = 50,000 years Figure 6-6. Distribution of the salinity at the repository level (Z = -410 m) at 1000, 10,000 and 50,000 years after the start of the disposal operations for model variants 2009SH and 2011HE (Löfman & Karvonen 2012 and Performance Assessment, Chapters 6 and 7). Model variant 2011HE corresponds to the DFN models used in the safety case. salinity at the repository depth and, specifically, areas with a low salinity develop with time. This model variant corresponds to the DFN models used in the safety case.

187 142 As the disturbances caused by repository construction cease, the groundwater composition will stabilise and the variation seen during the operational period will diminish. The few local values that were outside the target value range will return within the range in a relatively short time. At repository depth, the ph will remain close to 7.5 and reducing conditions prevail. During this time period, as a result of the infiltration of meteoric water at a slow, nearly constant rate, a decreasing trend in salinity, chloride and total charge equivalent of cations is expected. These values are, however expected to stay within the limits of the target ranges. The groundwater flow and transport modelling results indicate the possibility that a few canister positions may experience dilute conditions immediately after closure. The sulphide concentrations in the groundwaters after the post-closure period up to 10,000 years are expected to recover towards the steady state conditions, and it is expected that the initially controlling amorphous iron sulphide phases will successively evolve towards more crystalline iron sulphide phases with a lower solubility. However, as noted in Section a pessimistic sulphide concentration of 3 mg/l is adopted for use in subsequent analyses of canister corrosion. Thermal evolution of near field The ability of the buffer to transfer decay heat from the canister to the rock will remain sufficient to ensure the requirement of the maximum buffer temperature of 100 ºC is respected regardless of the presence of air-filled gaps and uncertainties in the mineralogical composition of the buffer. The better the buffer can transfer the decay heat from the canister to the host rock, the lower the canister surface temperature will remain. Mechanical evolution of the rock After the excavation and operational period and closure of the disposal facility, the rock stresses in the near field will be affected by the swelling of the buffer and backfill and by the thermal load from the spent nuclear fuel. There is a possibility of reactivation of fractures and rock damage, most notably thermally induced spalling, which may change the hydraulic properties of the near-field rock and thereby affect the target properties concerning limited groundwater flow and high transport resistance in the vicinity of the deposition holes. These factors are discussed above. Over time the thermal load will decrease and stable conditions will be reached. No performance targets will be violated due to the mechanical evolution of the rock after closure. Mechanical and hydraulic evolution of the buffer and backfill Groundwater flowing into the repository leads to saturation and swelling of the buffer and backfill. The time to reach full saturation in the buffer is calculated as a few tens of years to several thousands of years, depending on the local hydraulic conditions. Initial differences in the density and swelling pressure in the buffer and backfill will be evened out by homogenisation, although some heterogeneity will remain. Homogenisation is a process that is not completely understood and the development of numerical models will be continued. However, experiments and numerical assessments (THM-modelling) show that homogenisation takes place in the buffer-pellet-rock interface. Homogenisation has also been shown to take place in the backfill.

188 143 Initially, the low water content in the buffer will limit the microbial activity within it. Later on, at full buffer saturation, the high density will limit microbial activity. At the same time the buffer is impermeable enough to limit the mass flow to and from the canister and thereby limit transport of harmful substances, radionuclides and colloids. Calculations show that expansion of the buffer into the backfill and the changes in the density of the buffer will not be sufficient to threaten the performance targets for the buffer and backfill (i.e. a sufficiently high density will be maintained). Geochemical evolution of the buffer The complex thermo-hydro-mechanical-chemical (THMC) evolution during the thermal period will lead to geochemical changes in the buffer, but these will have limited impact on the performance targets. After saturation and development of the full swelling capacity, the changes will be much more moderate and constrained by diffusive processes. The increased temperatures in the buffer will induce no or only minor montmorillonite transformation (maximum 1 %) and very limited masses of newly cementing material (< 2 % by volume). The impact of cementitious leachates on montmorillonite transformation and porewater chemistry during the temperate period has been assessed based on the amounts of cementitious materials used in the repository. This effect has been estimated to be negligible (Section in Performance Assessment). Initially, the production of sulphide via microbial processes in the buffer will be inhibited by the low water content. After saturation, microbial activity will be restricted by the high buffer density. According to the results of groundwater flow modelling, the buffer at some deposition holes could potentially be affected by dilute waters and chemical erosion for a short period of time during the operational period and soon after closure. In these models, the infiltrating water is assumed to have zero salinity and no reactions in the overburden or along recharge path likely to increase the ionic strength of the infiltrating water are taken into account. This case is assessed together with cases suggesting dilute conditions and buffer erosion during a glacial cycle (Section 7.1 in Performance Assessment). Geochemical evolution of the backfill The evolution of porewater chemistry in the backfill will be similar to that in the buffer, but will be much less affected by the heat from the spent nuclear fuel (Section in Performance Assessment). Thus, any montmorillonite alteration and cementation due to thermally-induced changes will be negligible in the backfill. With regard to disturbances, the following conclusions can be drawn. a) The degradation of cement materials in the deposition tunnel end plug contacting the backfill will not affect the fulfilment of the performance targets of backfill during the temperate period or afterwards. Disturbances due to leachates from

189 144 cementitious materials will diminish in general and also locally due to the lower concentrations of the alkaline species in the leachates. b) The corrosion of iron from construction materials will have an insignificant impact on the performance targets of the backfill. c) The large sulphate pool in the backfill is a potential source for microbial sulphide production. In view of the large uncertainties related to backfill homogenisation and microbial activity in the boundary areas, the sulphide fluxes that may affect the canister can only be assessed by a bounding analysis. In the case of insufficient homogenisation and areas of lower density (for example in the interface area with the rock), sulphide may be produced by sulphate reducing bacteria. For the long time perspectives considered, the sulphide formed is not expected to rise above the sulphide levels calculated for mackinawite equilibrium ( mg/l). For any sulphide formed and even for the very pessimistic assumption that all of the sulphate will eventually be reduced to sulphide, the main processes attenuating the sulphide flux to the canister are slow diffusion transport, the precipitation of iron sulphide and the advective loss to the rock mass. d) In the case of good homogenisation, the high swelling pressure (high density) and small pore size will effectively restrict microbial activity and the conditions in the backfill will be similar to those in the (intact) buffer. If, however, low density areas should persist, then significant sulphate reduction cannot be ruled out, and thus it is considered in the canister corrosion analysis. Mechanical, hydraulic and geochemical evolution of the closure There are no major uncertainties in the evolution of the closure components during the first 10,000 years after closure. Even if it is assumed that the hydraulic plugs will become degraded, and some of the materials such as clays, aggregates and mixtures of these may be eroded or suffer settlement, this is expected to be a limited effect and no continuous preferential paths are expected to be formed. Therefore, at depth, transport through closure components will be dominated by diffusion during the first 10,000 years. Canister corrosion Sulphide is the main copper corrosion agent after all oxygen has been consumed. Microbially produced sulphide in the buffer is negligible in this period; sulphide supply from the backfill is limited by the precipitation of iron sulphide and losses to the rock mass, hence, the main source of sulphide is expected to be groundwater. Quantitative corrosion calculations coupled with groundwater flow modelling have been carried out (see Section 6.3 in this report and Section 7.7 Performance Assessment). These calculations also take into account the possibility of early buffer erosion due to low salinity, as mentioned above. Microbially produced sulphide in the buffer or in the backfill is considered negligible during this phase. The calculations show that total corrosion depth will be negligible during the first 10,000 years. The initially intact canisters will remain intact for all conceivable loads that could occur during the first 10,000 years (see below) and thus the spent nuclear fuel will remain contained within the canister.

190 145 Mechanical loading on canister During the temperate period, the canister(s) will remain intact, i.e. meet all its performance targets, for all conceivable loads (e.g. isostatic and uneven swelling pressure and groundwater pressure) that could occur during this period Fulfilment of performance targets and target properties As discussed above, for the expected evolution of conditions, during the post-closure period up to 10,000 years after present, the properties of the EBS and host rock will conform to the performance targets and target properties with a high degree of confidence. The few possible exceptions on their own do not threaten the integrity of the canisters. All relevant FEPs and FEP interactions have been evaluated and considered in reaching this conclusion. The key features and processes include: residual heat transfers local to individual containers and throughout the repository; heterogeneity of host rock and hence of groundwater flows and groundwater composition; long-term infiltration of meteoric waters; continued buffer and backfill swelling and homogenisation, piping and erosion; thermal and chemical reactions in the buffer and backfill (montmorillonite transformation), alteration of accessory minerals; localised and generalised corrosion of the copper overpack, due to sulphide. Groundwater modelling provides a quantitative evaluation of coupled thermal, hydraulic and chemical processes that accounts for the heterogeneity of host rock properties. There are some possibilities for incidental deviations, although none of these are expected. These incidental deviations are described below. As found for the operational period, it cannot be excluded that a few deposition holes might experience a higher post-closure flow rate or a lower transport resistance than the target values. Immediately after closure, modelling results indicate the possibility that a few canister positions may experience dilute conditions such that chemical erosion of the buffer could be possible. This result is considered to be due to simplified and pessimistic model assumptions and does not reflect the overall understanding of the likely future hydrogeochemical evolution of the site. Homogenisation of the buffer and backfill should ensure a sufficiently high density to restrict microbial activity; the conditions in the backfill will be similar to those in the buffer. If, however, low-density areas should persist in the backfill, then sulphate reduction to sulphide cannot be ruled out, and this is considered in the canister corrosion analysis. Of the above incidental deviations, none gives any possibility for release of radionuclides on its own. Even combining the pessimistic assumption on sulphide

191 146 (3 mg/l) in groundwater, microbial reduction of sulphate to sulphide in the localised parts of the backfill, and chemical erosion of the buffer in a few canister positions, the calculated corrosion depth is not enough to lead to canister failure in the 10,000 year time window. The case of a canister with an initial penetrating defect can be further affected by some of the above incidental deviations. These combinations are carried forward to be considered in the formulation of scenarios and assessment of radionuclide releases. 6.3 Beyond 10,000 years during repeated glacial cycles In the long term, i.e. over the next hundred thousand years or so, major climatic changes are expected to occur, being part of a future glacial cycle. According to the climate evolution described in Section 4.4, the current temperate period is assumed to last for the next 50,000 years. This is followed by a new cycle with characteristics based on the most recent glacial cycle. This implies alternating permafrost, warmer periods, permafrost and glaciations up to about 170,000 years after present (170 ka AP), with three glaciations, i.e. times during which the site is covered by an ice sheet between about 90 and 155 ka AP. Eight glacial cycles are postulated to occur in the next one million years. It is assumed that the cycles will include similar varying conditions, and similar loads will be imposed. After each glacial cycle a groundwater composition similar to that found at present day is expected to re-establish, although at repository depth the changes will in any case be minor and localised. The impact on the fulfilment of the performance targets and target properties can thus be assessed by considering the consequences of repeated loads similar to those assessed for the next glacial cycle. Key effects are: global sea-level fall, drier cold conditions and permafrost (ground freezing) and ice sheet formation and retreat, leading to changes in groundwater flows and composition; and mechanical loading and unloading of the crust due to ice-sheet growth and retreat, leading to rock stress changes and an increased likelihood of earthquakes during the retreat phase. These changes affect the mechanical and thermal evolution of the EBS and host rock Repository system evolution and evaluation of performance Hydraulic and geochemical evolution of geosphere Löfman & Karvonen (2012) discuss the groundwater flow and salinity evolution during the continued temperate period until 50,000 years AP, under permafrost conditions and during an ice-sheet retreat. The modelling has been carried out for selected time periods from the reference climate evolution to represent permafrost and ice-sheet conditions. Different model variants have also been used in the modelling. Further groundwater flow modelling and reactive transport modelling during the ice-sheet retreat is presented by Hartley et al. (2013b) and Trinchero et al. (2013).

192 147 During the continued temperate climate from 10,000 years AP until 50,000 years AP, a slight increase is expected in the groundwater flow rates in the upper part of the bedrock down to approximately 300 m depth. This increase is related to the increase of hydraulic head due to the changes in the surface environment. Following the retreat of the coastline, peatlands are expected to develop in the low-lying areas and the hydraulic head in these areas will increase. The difference in hydraulic head between the area above the repository and the discharge areas around the present island changes from 12 m (10,000 years AP) to 14 m (50,000 years AP). Flow rates at the repository depth will, however, not be significantly affected. The groundwater flow modelling under permafrost conditions has been carried out for two representative distinct periods of the last glacial cycle (Figure 6-7). Both these periods last about 10,000 years. During period 1, the permafrost reaches a depth of approximately 80 m, and, during period 2, permafrost reaches around 300 metres depth. The assumed permafrost depths are based on Hartikainen (2013). At the beginning of the permafrost periods, the hydraulic conditions are taken as those at 10,000 years AP due to the better reliability of the data and models for this period. Modelling cases with and without taliks have been defined. It has been assumed that taliks may form in the current sea area north, northwest and southwest of Olkiluoto Island (see Haapanen et al. 2010). Under permafrost conditions, the hydraulic conductivity of the rock is reduced by several orders of magnitude (see Löfman & Karvonen 2012) and the infiltration is very low. Consequently, the groundwater flow is significantly reduced (see Figure 6-7). Modelling groundwater flow and salinity evolution during the retreat of the ice sheet comprised three cases with an ice margin staying at Olkiluoto for 1000 years, but at different locations with respect to the repository, and a case with a constantly retreating ice sheet. In all simulations, the ice front was initially located at a position such that the ice sheet overlay the whole of Olkiluoto Island. The retreating ice front then moved to case-specific locations at a rate of 200 m/year. The salinity at the initial state for the simulations of ice-sheet retreat period was assumed to be that at the end of permafrost period 1 with taliks. During the ice sheet retreat, the flow rates through the repository volume depend on the location of the ice margin with respect to the repository (see Figure 6-8). While the repository is still below the ice sheet (although the ice margin is close), the flow rates are increased by a factor of 4 to 7 compared with the situation at the end of the temperate climate or with the natural state before construction of the ONKALO.. In the subsequent analyses, it has been assumed that the flow rates are 10 times higher during the ice-sheet retreat than during the temperate time windows. The flow direction below the ice sheet is mainly downwards and thus, during this time, the flow paths entering the repository originate mainly from areas below the ice sheet. As the ice sheet passes the site, the main flow direction is upwards and the flow rates reduce as the distance to the ice margin increases. The site is likely to be submerged (below sea level) after the retreat of an ice sheet. During the submerged period, when the ice margin is not in the vicinity of the site, flow rates at the repository depth are lower than during the temperate period as the only driving force is density variation.

193 148 Period 1 Period 2 Figure 6-7. Total flow rate to the reference volume under the two permafrost periods considered (Löfman & Karvonen 2012). For explanation of 2009SH, 2011SH and 2011HE, see Figure 6-1. The permafrost development was from Hartikainen (2013), 1D results were used in the groundwater flow modelling. Taken from Performance Assessment (Chapter 7).

194 149 Figure 6-8. Total flow rate into the reference volume at the time of the ice-sheet retreat; above mobile ice sheet retreating with a velocity of 200 m/year and below the immobile ice sheet located northwest of the repository (Löfman & Karvonen 2012). 2009SH model variant 1, 2011SH model variant 2, 2011HE model variant 3. Taken from Performance Assessment (Chapter 7). During the continued temperate period, the infiltration of meteoric water at a slow and nearly constant rate results in a decreasing trend in salinity. The modelling results show that, towards the end of this period, a few percent of the canister positions may experience dilute conditions. As a result of the significantly reduced groundwater flow during the permafrost periods, the groundwater salinities remain at the level prevailing before the onset of the permafrost. Dilute conditions may also be experienced during the ice-sheet retreat phase, but the estimate of the number of such positions is strongly dependent on the duration of the melt water intrusion and especially on the modelling concept of the interaction between the fracture water and the rock matrix. Set against

195 150 this, there is no evidence that fresh meltwater has reached repository depth at Olkiluoto during the last glacial cycle or the previous ones. However, these uncertainties are considered in scenario development, and lead to cases in which some deposition hole locations are assumed to experience dilute conditions during the initial extended temperate period and in association with glacial (ice-sheet margin) conditions. The number of positions affected is influenced by the duration of the meltwater intrusion. Similarly to the salinity results, other key geochemical properties (i.e. ph, Eh, Cl concentration, total charge equivalent of cations, sulphur and iron species) are also expected to stay within the limits established for the target properties during ice-sheet retreat and melting for most of the deposition holes. Oxygen will be consumed within short distances along the flow path and thus does not reach the repository level. Freezing/thawing of buffer and backfill The potential for freezing of the buffer or the deposition tunnel backfill is not an issue since permafrost will not reach repository depth (Hartikainen 2006, 2013). Even if the freezing front were to reach the repository depth, the materials, and design selected for the buffer and backfill would withstand the freeze/thaw cycles without damage to their safety functions (Schatz & Martikainen 2010, 2013). The stability of clay materials against freeze/thaw cycles also implies that these performance targets would also be upheld for more extreme climate evolutions. Geochemical evolution of buffer and backfill The evolution of porewater salinities in the buffer and backfill will follow those in the surrounding groundwaters, which will remain within the required performance target ranges, except perhaps during short times within the ice retreat and melting period. Under these conditions, dilute groundwater conditions may cause some chemical erosion of buffer and backfill and local lowering of density. In these deposition holes and tunnel sections with lower density, sulphate reduction may occur. However, the sulphide concentration in solution is still limited by the availability of iron in the materials and the surrounding rock, which results in precipitation of iron sulphides, keeping the sulphide concentration below 3 mg/l. Ongoing degradation of cementitious materials will gradually release less aggressive leachates, whose effects on the clays in the near field will generally be very limited. Changing hydraulic conditions and freezing induced by glaciation effects however may locally increase the release of cement leachates in closure components in the upper part of the rock. The long-term stability of montmorillonite under repository conditions is difficult to assess quantitatively because of lack of theoretical and experimental knowledge. Nonetheless, kinetic constraints and especially observations from natural systems indicate long-term stability of montmorillonite at low temperatures over a large range of geochemical conditions. Chemical erosion of buffer and backfill Chemical erosion of buffer and backfill in some deposition holes and deposition tunnels due to the potential occurrence of dilute groundwater cannot currently be ruled out for short times during an ice retreat and melting period. With the reference assumptions on

196 151 groundwater flow (a selected realisation of the DFN flow model) and evolving groundwater composition, one canister position is calculated to undergo buffer erosion during the first glacial cycle to an extent that advective conditions arise. This calculation should be seen as illustrative, being based on only a single realisation of the DFN groundwater flow model. An analysis of statistical distributions of flow-related parameter values between canister positions shows that measures such as the mean and 90th percentile vary little between DFN realisations. However, the number of canister positions experiencing advective conditions is determined by the tails of these distributions, and is therefore subject to more uncertainty. Taking a more cautious view on this and other uncertainties, buffer erosion might result in advective conditions in a few canister positions. The modelling is subject to a number of uncertain including: whether sufficiently dilute conditions are attained and for how long; the groundwater flow distribution; the extent of chemical erosion; threshold values for buffer and backfill loss before advective conditions are attained; the implications of application of RSC criteria. The consequences, if any for canister corrosion, are considered in Formulation of Radionuclide Release Scenarios. Evolution of the closure components After 10,000 years, the backfill in the central tunnels will be completely saturated and keep its safety functions regardless of the consequences of climate evolution. For the upper parts of the closure, the following is concluded. It cannot be excluded that the backfill in parts of the access tunnel will lose its clay components due to chemical erosion over the long time scale considered. However, this is not judged to jeopardise the overall safety functions of the closure backfill in particular and of the closure components, as a whole. Degradation of closure plugs is uncertain, but the swelling clays used in the lower parts of the tunnels and shafts will ensure sufficient isolation capacity of the sealing structures. Freezing/thawing of the closure components would not impair their performance relative to the closure performance targets. The access tunnel and shafts between depths of 200 and 300 m backfilled with an in situ compacted swelling clayaggregate mixture may, in the far future, be subject to freeze/thaw cycles, but this will have no major implications for the performance of the material. The material filling the access tunnel and shafts above 200 m depth is in situ compacted crushed rock. If frost heave develops, it will be of minor consequence, if the crushed rock material is selected appropriately. Glacial erosion may have an effect on materials in the upper part of the disposal facility, but the erosion rate, even during glacial cycles is so slow (average of 8 mm in 100 years; Okko 1964) that it would take several millions of years to erode the upper plugs and the material underneath.

197 152 Rock mechanics and mechanical impacts on the canister Large earthquakes may occur in connection with ice-sheet retreat. There is the possibility that shear displacements exceeding 5 cm could cause canister failure in the context of such earthquakes. Although, by locating the deposition holes away from large deformation zones and avoiding large fracture intersections in deposition holes, it is possible to decrease the risk of canister failure due to an earthquake, it is estimated that few tens of canisters could be in positions such that they could potentially fail in such an event. On the other hand, the average annual probability of an earthquake large enough potentially to lead to canister failure is estimated to be low, in the order of 10-7, given that there are around five fault zones within and around the area of the repository that could host such an earthquake. Thus, during the first glacial cycle, the probability of occurrence of such an earthquake is low and it is very likely that the canisters will remain intact, i.e. meet all its performance targets, for all conceivable mechanical loads. For the next four glacial cycles, the average annual probability of an earthquake large enough potentially to lead to canister failure is of a similar order of magnitude as during the first glacial cycle, even if an earthquake were to occur during the first cycle. This is because, with a low tectonic strain rate, it takes a long time to compensate for the loss of stored strain energy after the stress relaxation due to a single earthquake. Furthermore, as fracture propagation is limited, even if another earthquake were to occur, the same sub-set of fractures is likely to be reactivated. This, together with the low annual probability of large earthquakes, keeps the annual probability of canister failure low. However, over a one million year time frame, the possibility of some canister failures by this mechanism cannot be excluded. In case of a small rock movement, the toughness, plasticity, creep and relaxation properties of copper ensure that the overpack (now in contact with the insert) does not break in spite of the shearing force causing deformation of the canister. Canister corrosion If the buffer is intact, a sulphide concentration of more than mg/l is necessary to completely corrode the copper shell thickness of 49 mm (or 35 mm considering the minimum required copper thickness) in one million years for the most unfavourably located deposition hole. This concentration is more than two orders of magnitude higher than the pessimistic bounding estimate for sulphide concentration in groundwater (3 mg/l). Figure 6-9 shows the distribution of corrosion depth over a one million year time period in all deposition holes satisfying the inflow RSC criteria, and also with no application of RSC, for three cases. The first two consider diffusion of sulphide directly across the buffer from a fracture intersecting the deposition hole: a) a groundwater sulphide concentration of 3 mg/l and the distribution of flow from groundwater flow modelling assuming no hydraulically significant damaged zone around the deposition hole. In this case, most canisters experience very limited corrosion (less than 0.1 mm of corroded copper) over one million years even if the RSC criteria are not applied.

198 153 (a) (b) (c ) Figure 6-9. Number of canister positions as a function of corrosion depth over 1 Ma, with and without the RSC applied, in the case of direct diffusion of sulphide across the buffer from a fracture intersecting the deposition hole with (a) no hydraulically significant rock damage around the deposition hole (b) increased flow around the deposition hole due to rock damage, and in the case of (c) downward diffusion of sulphide from the deposition tunnel. In this case also the limiting corrosion depth for high groundwater flow in the deposition tunnel is given. For all cases the assumed groundwater sulphide concentration is 3 mg/l.

199 154 b) the same groundwater sulphide concentration but an increased flow around the deposition hole due to rock damage. In this case, corrosion depths are somewhat higher (but still very small, in the order of 0.2 mm); the application of RSC has a limited effect in this case. c) the third case assumes a downward diffusion of sulphide from the deposition tunnel, again with a 3 mg/l sulphide concentration. In this case, the distribution of corrosion depths is markedly higher, but still not sufficient to lead to canister failures. The results show that the overall corrosion depth will not exceed a few tenths of a millimetre even over one million years. Thus, if the buffer performs as designed, no canister failures are expected even with high sulphide concentrations. Furthermore, even if the buffer is affected by chemical erosion, few if any canister failures due to corrosion are expected during the first glacial cycle, as long as conditions otherwise correspond to the expected evolution (i.e. performance targets and target properties are met). The calculated rate of corrosion and the calculated number of canister failures in these circumstances depends on the assumptions made about groundwater flow and composition, corrosion area, fracture apertures, the rate of buffer erosion and the possibility of locally thinner parts of the copper overpack. Cautiously assuming a sulphide concentration of 3 mg/l in the groundwater, but with realistic assumptions concerning these other factors, chemical erosion of the buffer and subsequent corrosion by sulphide is calculated to lead to no canister failures within the first glacial cycle, and 4 5 failures in the million year time frame. Based on more cautious assumptions, around 3 canister failures are calculated to occur within the first glacial cycle, and a few tens of failures in the million year time frame. No canister corrosion failures are expected during the first glacial cycle, even if the buffer is chemically eroded, as long as the conditions correspond otherwise to the expected evolution (i.e. performance targets and target properties are met). With pessimistic assumptions concerning the intact wall thickness (35 mm) of the canister, the corrosion area, fracture aperture, high flows and duration of dilute conditions, chemical erosion of the buffer in a small number of boreholes and subsequent corrosion by sulphide might lead to a few canister failures in the time frame of one million years (assuming sulphide concentration of 3 mg/l in groundwater). The number of canister failures depends on the assumptions about groundwater flows, fracture apertures and buffer erosion modelling. Thus, most of the canisters will remain intact for the time period up to one million years and provide complete containment of the spent nuclear fuel. As discussed above, for most of the deposition holes the performance targets for the buffer, and also the target properties of rock, are upheld. Incidental deviations for some deposition holes may result in dilute conditions causing chemical erosion of the buffer, allowing advective conditions that may lead to a few canister failures by corrosion on a timescale of several hundred thousand years. Copper corrosion by water in oxygen-free conditions is still under investigation to understand some of the results published in the literature (Feature, Events and Processes, 4.2.5).

200 155 Subcriticality In the long term, after canister failure, the probability of criticality is also very low because there are no credible mechanisms that could cause the redistribution of the fissile material into a critical configuration either inside or outside the canister. Although long-term criticality is not considered in the formulation of release scenarios, this issue is still under investigation Fulfilment of performance targets and target properties As discussed above, for the expected evolution of conditions during the next glacial cycle, i.e. more than 100,000 years into the future, the properties of the EBS and host rock will conform to the performance targets and target properties with some incidental deviations over most of the repository and most canister positions with a high degree of confidence, and most of the canisters will remain intact. Even in the one million year time frame, the components mostly still meet their performance targets. The number of deposition holes with incidental deviations with respect to groundwater flow target properties, salinity and sulphide levels, and potential for shear displacements, are essentially the same as after the first glacial cycle. All relevant FEPs and FEP interactions have been evaluated and considered in reaching this conclusion. The key features and processes include: changes in groundwater flows due to changes in sea level, temperature and precipitation, permafrost and ice sheets; long-term infiltration of meteoric water and inflows of glacial waters; freezing and thawing of closure components; chemical erosion of buffer and backfill; stress redistribution and earthquakes related to ice-sheet retreat; long-term generalised corrosion of the copper overpack, due to sulphide. Groundwater flow modelling provides a quantitative evaluation of coupled thermal, hydraulic and chemical processes, including the impacts of heterogeneity of host rock properties. There are some possibilities for incidental deviations from the target properties related to the flow conditions, especially during the high-flow conditions related to ice-sheet retreat; although none of these are certain and these deviations can only affect a few canister positions. Dilute conditions may be experienced in some deposition holes during ice-sheet retreat, but the number of such positions depends on the duration of the melt water intrusion and on the modelling assumptions on the diffusional mass exchange between the fracture water and the rock matrix, dilute conditions are possible and more widespread, if matrix diffusion is either not accounted for or is assumed to be limited only to a short distance from the flowing fractures. In spite of its geological stability, a large earthquake could potentially occur near the site, especially in connection with ice-sheet retreat. The number of deposition holes that could experience shear displacements on intersecting fractures large enough to cause

201 156 canister failure in such an event is kept low by locating the deposition holes away from the large deformation zones and avoiding large fracture intersections in the deposition holes. Nevertheless, and although the average annual probability of an earthquake large enough potentially to lead to canister failure is estimated to be low, in the order of 10-7, the possibility of a limited number of canister failures by this mechanism cannot be excluded over a one million year time frame. As long as transport of corrodants through the buffer remains diffusion dominated, no canister failures by corrosion are expected within 1 million years, even in the least favourable canister locations and even with pessimistic assumptions regarding uncertainties in groundwater flow and diffusion of sulphide from the backfill. However, chemical erosion of buffer due to the occurrence of dilute groundwater cannot be neglected. The calculated rate of corrosion and the calculated number of canister failures in the event that erosion leads to advective conditions in the buffer depends on the assumptions made about groundwater flow and composition, corrosion area, fracture apertures, the rate of buffer erosion and the possibility of locally thinner parts of the copper overpack. Cautiously assuming a sulphide concentration of 3 mg/l in the groundwater, but with realistic assumptions concerning these other factors, chemical erosion of the buffer and subsequent corrosion by sulphide is calculated to lead to no canister failures within the first glacial cycle, and 4 5 failures in the million year time frame. 6.4 Summary statement of performance and uncertainties The previous sections have summarised how the properties of the EBS and host rock are expected to conform to the performance requirements over more than 100,000 years. Thus no radionuclide releases are expected over more than 100,000 years for the expected evolution of conditions, primarily because, protected by the buffer, the copper canisters are designed and expected to withstand all likely events and conditions over this period. A full exploration of uncertainties and variability in initial conditions of the EBS and host rock, and site evolution, has identified some events and conditions that could lead to incidental deviations from performance targets and target properties (Table 6-1). The importance of these has been assessed using different release scenarios.

202 157 Table 6-1. Summary of incidental deviations from performance targets and target properties that may occur and are relevant in each time window. Deviations Up to closure of the disposal facility Up to 10,000 years Possibility of an initial penetrating defect in one or a few canisters. Higher flow rate or lower transport resistance than the target ranges for a few deposition holes. Groundwater composition outside the target ranges for a short time during operation and soon after closure for a few deposition holes. Low density areas in the backfill where sulphate reduction to sulphide cannot be ruled out. Erosion of buffer in some deposition holes due to long-term infiltration of meteoric water or dilute glacial meltwater. Canister failure by corrosion due to unfavourable groundwater conditions and buffer erosion. Canister failure due to shear displacements in fractures during ice-sheet retreat During repeated glacial cycles

203 158

204 159 7 FORMULATION OF RADIONUCLIDE RELEASE SCENARIOS AND CALCULATION CASES A scenario represents one time history of conditions (hereafter called a line of evolution) or more than one time history (lines of evolution). The performance of the repository system and its components has been analysed taking into account the expected lines of evolution and the uncertainties involved (Chapter 6). Although in Performance Assessment it is shown that no radionuclide releases are expected during the first 10,000 years after emplacement and not even within 100,000 years, uncertainties in the initial state of the barriers, groundwater evolution and the occurrence of unlikely events have been examined by scoping calculations in Performance Assessment. If the scoping calculations show that the line(s) of evolution may lead to the failure of one or more barriers (i.e. the failure of one or more safety functions) such that radionuclides could be released, then a set of calculation cases (reference case, sensitivity cases, what if cases) is defined within the scenarios to analyse the impact of the potential radionuclide releases. This is documented in Formulation of Radionuclide Release Scenarios and summarised in this chapter. The analysis of these scenarios and calculation cases is documented in Assessment of Radionuclide Release Scenarios for the Repository System and Biosphere Assessment, and summarised in Chapter 8 of this report. 7.1 Lines of evolution framing the scenarios and scenario formulation As described in Section 2.3.6, scenarios for radionuclide release calculations are based on defining lines of evolution for the repository system and surface environment, and consistent with the definition of scenario types Base scenario, Variant scenarios and Disturbance scenarios given in Guide YVL D.5. For the TURVA-2012 safety case, lines of evolution of the disposal system are defined following the timeline given by the climatic evolution (see Section 4.4) selected based on the recommendations by Pimenoff et al. (2011). The climatic evolution defines the time windows in which climate-driven processes may operate. Processes internal to the disposal system (whether or not driven by external events and associated changes in external conditions) are also taken into account. The lines of evolution that comprise the expected climatic evolution and frame the disposal system evolution are as follows. 1. The climatic line of evolution takes into account a temperate period (i.e. boreal climate) similar to the current one, which is assumed to last from today until about 50,000 years after present according to Table 3 in Pimenoff et al. (2011). Thereafter, effects of present-era human effects are assumed to have subsided and a return to glacial-interglacial cycling is expected; this is represented by assuming a repetition of the Weichselian glacial-interglacial cycle (which lasted about 120,000 years). This is a stylised representation, since in the last million years none of the glacial cycles has been a repetition of another, but it is a reasonable and evidencebased choice. There could be shorter or longer permafrost periods and less or more extensive ice cover. Very pessimistic climate conditions would be required, however, for permafrost to reach repository depth (Hartikainen 2013). There are

205 160 uncertainties in the timing and duration of periods of permafrost, and therefore only specific time windows of permafrost occurrence have been selected for detailed studies. The most reliably characterised ice-sheet retreat period at the end of the Weichselian was also selected for detailed study. 2. The evolution of the geosphere 22 and surface environment includes the hydrogeological evolution (groundwater and surface water; coupled to thermal and mechanical evolution in the context of groundwater evolution); the evolution of surface water gives the boundary conditions for the hydrogeological evolution of groundwater (see Löfman & Karvonen 2012 and Sections 5.1, 6.1 and 7.1 in Performance Assessment). The hydrogeochemical evolution, i.e. variations in salinity and groundwater composition due to gradual dilution with meteoric water and possible intrusion of dilute glacial meltwaters, is coupled to the hydrogeological evolution. The rock mechanical evolution is coupled to the thermal evolution and also to stress changes due to glacial loading. All these couplings are discussed in Performance Assessment. 3. The evolution of the EBS (i.e. canister, buffer, backfill and closure components) is coupled to thermal, climatic, hydrogeological, hydrogeochemical and mechanical evolution for the same time windows. The evolution of the canister (with no initial penetrating defects) 23 is coupled to buffer and backfill evolution and consequently to climatic and geosphere evolution. Again, these couplings are discussed in Performance Assessment The link to scenario hierarchy The base scenario, as defined in Section 2.3.6, includes the expected, or most likely, lines of evolution of the repository system taking into account external conditions (i.e. climate evolution), internal phenomena, and human actions. For the repository system this includes the incidental deviation whereby one or a few canisters with an initial penetrating defect are emplaced in the repository, giving rise to releases of radionuclides. All other canisters are manufactured and emplaced according to design. Surface environment scenarios are formulated by adopting a concept of credible lines of evolution. This is based on the expected line of climatic evolution and physical changes in the surface environment, e.g. sea-level change and development of natural ecosystems, but employs a stylised representation of future human activities based on present-day habits, e.g. regarding land use. Evolution lines within variant scenarios consider the diminished performance of the safety functions of the canister and/or the combined effects of the reduced performance of more than one safety function of the other barriers, still within the broadly expected conditions. Variant scenarios for the surface environment consider alternative credible 22 The evolution of the geosphere at the Olkiluoto site is discussed (see Sections 5.1, 6.1, and 7.1 in Performance Assessment ) for the time window of the ongoing temperate period (boreal climate), for selected time windows of cold and dry climate (permafrost), and for selected time windows of ice-sheet advance and retreat (especially ice-sheet retreat, which tends to be more threatening to the isolation properties of the repository). 23 Note that the evolution of one canister or more with an initial undetected penetrating defect will also be coupled to the evolution of the system as described above.

206 161 lines of evolution arising from reasonable variations of key FEPs, such as alternative discharge locations, sea-level changes, land uses and human habits. Lines of evolution including unlikely features, events and processes are considered within disturbance scenarios and consider both unlikely natural phenomena (e.g. large earthquakes, intrusion of dilute melt waters to repository depth to the extent that the buffer becomes significantly eroded) and unlikely phenomena related to human actions (e.g. inadvertent human intrusion and unlikey dietary profiles) Process for identification of scenarios and cases Repository system Posiva s approach to scenario formulation for the repository system follows a top-down approach. This includes first identifying the safety functions, then considering the effect of single FEPs or combination of FEPs on these functions, and also the effect of uncertainties within the expected lines of evolution. The regulatory framework is taken into account; it is prescriptive in terminology and definitions. Thus: The safety functions for each of the repository system components are defined and a range of values (performance targets and target properties) is given whenever possible (see Section 2.2 and Design Basis). FEPs that could adversely affect one or more safety functions at a given time or place or under specific conditions within the repository are identified (i.e. FEPs that are scenario drivers within the evolution of the repository system in time and space (see Performance Assessment). The effects of uncertainties in the expected evolution of the repository system are taken into account (see Performance Assessment). Lines of evolution that describe the evolution of the repository system and ultimately lead to canister failure, form the basis for the definition of radionuclide release scenarios. Each line of evolution is then classified using STUK s scenario terminology (Section 2.3.6). For each of the scenarios, a set of calculation cases is defined to analyse the potential radiological impact. The calculation cases take into account uncertainties in the assumptions and data through variations in the models and parameter values. The most relevant evolution-related FEPs that may affect the safety functions of the repository system, and thus also affect migration-related FEPs, have been taken into account in analysing and describing the expected evolution (see Performance Assessment). Climate evolution (see Section 4.4) is the overarching FEP affecting the whole disposal system. In the expected evolution, the safety function of the canister(s) holds for hundreds of thousands of years (see Performance Assessment) and no releases would occur within the time window of the first several millennia. Nonetheless, according to regulations (Government Decree 736/2008), a base scenario needs to be defined with a high probability of causing radiation exposure, but of low consequences.

207 162 Therefore, the base scenario leading to radionuclide releases is defined as a scenario in which one or a few canisters with an initial penetrating defect are emplaced in the repository, while the safety functions of all the other canisters and the other repository system components are maintained as described in Performance Assessment. Variant scenarios are identified that consider plausible alternative assumptions for key FEPs that could affect safety functions and the release of radionuclides. Disturbance scenarios consider very unlikely events or processes that could lead to loss of safety functions and release of radionuclides. Finally, complementary cases investigate key uncertainties that help to give a better understanding of the modelled system or subsystems. The complementary cases are fully documented in Formulation of Radionuclide Release Scenarios and analysed in Assessment for Radionuclide Release Scenarios for the Repository System and Biosphere Assessment. The surface environment Formulation of scenarios for the surface environment must be consistent with the regulatory requirements, the methodology used in the formulation of scenarios for the repository system, and the current radiation protection systems for humans and the environment. Posiva s approach for the surface environment is somewhat different from that for the repository system, since the surface environment has no safety functions. The process for identifying scenarios and cases is as follows. Constraints on the scenarios arising from the regulatory framework are identified. The most important FEPs (denoted key scenario drivers) with respect to the evolution of the surface environment, fate of radionuclides in the surface environment and/or the radiation exposure of humans, plants and animals are identified. This work also comprises identifying other FEPs that affect the key scenario drivers, either in isolation or combined, and could induce changes in a timeline of evolution. One or several lines of evolution are defined that describe in a timeline the surface environment evolution from which one or more scenarios are formulated. One credible line of evolution is identified and used to formulate the Base Scenario for the surface environment. Variant scenarios are formulated, mainly by considering reasonable deviations from the lines of evolution underpinning the Base Scenario. Variant scenarios can include additional scenario drivers (FEPs) with a potentially significant effect on the fate of radionuclides in the surface environment and/or the radiation exposure of humans, plants and animals. Disturbance scenarios are formulated, mainly by identify unlikely FEPs or mainly by considering unlikely deviations from the lines of evolution underpinning the Base Scenario. Disturbance scenarios can include additional scenario drivers (FEPs) with a potentially significant effect on the fate of radionuclides in the surface environment and/or the radiation exposure of humans, plants and animals. A set of biosphere calculation cases is defined to analyse the surface environment scenarios. These cases take into account uncertainties in assumptions and models, and the uncertainties and variability in parameter values applied in the models.

208 Repository system scenarios Base scenario for the repository system Derivation of the base scenario Considering scenarios for radionuclide release, the canister (which provides the safety function of prolonged containment of the spent fuel) is the primary barrier since radionuclide releases may only occur if the canister has failed. Possible canister failure modes are generalised and localised corrosion, rock shear, and presence of an initial undetected defect (see Chapter 6). As discussed in detail in Performance Assessment, canister failure through corrosion is only possible at very long times in the future. When using pessimistic assumptions, a few canisters may fail within one million years after chemical erosion of the buffer leads to advective conditions and to enhanced corrosion. Canister failures through rock shear are only possible as a result of a large earthquake following rock stress changes due to glacial loading and unloading, and thus will only occur in the time frame of about one hundred thousand years or more. Although the average annual probability of an earthquake large enough potentially to lead to canister failure is estimated to be low, in the order of 10-7, the possibility of a limited number if canister failures by this mechanism cannot be excluded over a one million year time frame. It is expected that most of the canisters can be technically manufactured as designed without any initial penetrating defect, and any penetrating defect greater than about 0.5 mm diameter is likely to be detected by weld inspection and testing (Holmberg & Kuusela 2011, see also Section 3.5). The base scenario postulates that one or a few defective canisters are emplaced in the repository. The currently available data are insufficient, even when expert judgement is used, to make a reasonable estimate of the probability of emplacing a defective canister in the repository. However, with additional data on the welding process and continued development of the NDT process, it seems practicable in the future to show that the probability of more than one initially defective canister in the repository is less than one per cent. Thus, for the base scenario one canister with an undetected penetrating defect of 1 mm diameter is assumed in the Reference Case and in other cases addressing this canister failure mode. This is consistent with YVL Guide D.5, which states that the base scenario shall assume the performance targets for each safety function, taking account of incidental deviations from the target values. Thus, in the base scenario Reference Case, the undetected penetrating defect in one canister is the incidental deviation, which acts as the main driver, whereas the performance targets of most of the canisters and of all the other repository components are expected to hold as shown in the performance assessment. The likelihood and consequences of more than one defective canister being emplaced are, however, considered in the complementary analysis described in Section It is assumed that the deposition tunnels have been excavated, and deposition holes selected, successfully applying Rock Suitability Classification criteria. This implies that during the disposal system evolution: 1) there will be a very low probability of earthquakes leading to rock shear displacements along existing fractures or

209 164 discontinuities that could damage the engineered barrier system in most of the selected deposition holes; 2) the evolving hydrogeological, hydrogeochemical, and mechanical conditions will not impair the safety functions of the buffer, backfill, closure or host rock for the majority of the deposition holes. Since initial canister defects are uncorrelated to where the canister will be deposited, this means that for the Reference Case it can be assumed that the canister with initial defect will experience rock and engineered barrier systems with intact safety functions. The assumptions regarding the EBS and host rock are shown in Table 7-1. Nevertheless, there is substantial uncertainty in the radiological outcome of the base scenario arising from the location of the defective canister. To analyse the case, the canister position that may lead to radionuclide releases to a discharge location or locations in the surface environment within the first millennia have been selected from a chosen DFN realisation. This is valid since the values that would be assigned to cautiously selected flow-related parameters have been shown not to vary greatly between DFN realisations. Radiological impacts of multiple canister failures might or might not coincide spatially and temporally (be additive) depending on location and timing of release to the surface environment. Lines of evolution, key processes and uncertainties The climate evolution that encompasses the repository system evolution is the one presented in Chapter 4 in Formulation of Radionuclide Release Scenarios and summarised in Section 4.4 of this Synthesis. The evolution shows alternate temperate, permafrost and ice-sheet periods. For the first tens of thousands of years and prior to the first permafrost period at about 50,000 years after present, the hydrogeologically effective precipitation data derived for a climate evolution for a CO 2 concentration of 400 ppm or 280 ppm are very similar. Precipitation data are used for surface hydrology modelling that serves as a boundary condition for groundwater flow modelling (Löfman & Karvonen 2012). After 50,000 years permafrost may develop, but it is cautiously assumed to have no effects on the release rate of radionuclides or on in the release paths that could retard the transport of radionuclides to the surface. The repository system in the base scenario follows the expected evolution depicted in Chapters 5 to 8 in Performance Assessment in which the majority of the canisters, and the buffer, backfill, closure components and host rock, maintain their respective safety functions for the whole assessment period. It is assumed that the transport of radionuclides from the defective canister to the rock takes place through a fracture around the deposition hole, and that there is a damaged zone around the deposition hole. Given that the transport path from the defective canister to the near field is pessimistically assumed to become established within the first millennia, the groundwater type is selected according to current observations and modelling results (Löfman & Karvonen 2012), and thus is assumed to be brackish. As stated above, the position(s) of the defective canister in the repository have been cautiously selected from a chosen DFN realisation that takes into account the whole repository system (see Section 6.2 in Assessment of Radionuclide Release Scenarios for the Repository System).

210 165 Table 7-1. Assumptions for the base scenario for the repository system (note also the enveloping climate evolution discussed in Section 4.4). Bedrock Engineered Barrier System (EBS) Rock mass Groundwater Closure Deposition tunnel backfill Buffer Canister RSC criteria are applied successfully and target properties hold during the evolution. Limited advection or inflows to repository level. Favourable groundwater chemistry for the EBS and target properties for the groundwater chemistry hold during the evolution. Closure backfill and seals, including borehole seals are designed and emplaced according to requirements, and performance targets are fulfilled during the evolution. Deposition tunnel backfill and plugs are designed and emplaced according to requirements. The backfill performance targets are fulfilled during the evolution. The buffer is designed and emplaced according to requirements. The buffer performance targets are fulfilled during the evolution. Canisters are manufactured and emplaced according to design. As an incidental deviation it is assumed that one canister is present with an initial undetected penetrating defect of 1.0 mm diameter, which size will not change in time. Spent nuclear fuel and cladding Very low dissolution rate; no requirements or safety function in itself. Figure 7-1 illustrates the repository base scenario and the uncertainties related to the engineered and natural barriers and their safety functions addressed in the scenario. The uncertainties are related to: 1) location of the defective canister within the repository, and the corresponding flow-related transport parameters and discharge locations to the surface environment; 2) the time of establishment of the transport path ending in the selection of more than one time; 3) the speciation of several radionuclides as anions or cations in the near and far field (Wersin et al. 2013a and b, Hakanen et al. 2013) ending in the selection of more than one parameter (i.e. sorption, diffusivity) value for those radionuclides.

211 166 Figure 7-1. The repository base scenario and the uncertainties addressed related to the repository barriers and their safety functions Variant scenarios for the repository system The reduced performance of any single safety function(s) of any component other than the canister does not immediately give rise to canister failure and thus to radionuclide releases. However, the reduced performance of the buffer may subsequently affect radionuclide release and transport. The combined effect of the reduced performance of the canister and the buffer is assessed in two variant scenarios, where the loss of the safety function of the canister (initial penetrating defect, or failure by corrosion) is combined with the reduced performance of the buffer. Variant scenario 1 (VS1): Enlarging defect and degradation of the buffer In this scenario, one canister is assumed to have an initial penetrating defect of 1 mm diameter at the time of emplacement that, due to corrosion, will enlarge up to 10 mm on a timescale of 25,000 years. The degraded performance of the buffer is assumed to have arisen as a consequence of a process or combination of processes that are likely to occur within the first tens of thousands of years after emplacement, such as piping erosion, and/or montmorillonite transformation (e.g. due to heat-transfer-induced cementation, iron-clay interaction, and interaction with high ph water), which may lead to a reduced effective buffer thickness. Montmorillonite transformation would occur preferentially at the interface between the canister and buffer (if induced by heat transfer) or at the interface between buffer and

212 167 host rock. Piping and erosion would occur preferentially at the host rock/buffer interface. In either case, the whole thickness of bentonite would not be affected. It is assumed that the transport of radionuclides from the defective canister to the rock takes place though a fracture around the deposition hole, and that there is a damaged zone around the deposition hole, as in the base scenario. The uncertainties considered in this scenario are related to: 1) the composition of groundwater, also reflected in the selection of the sorption, diffusion and solubility values for the geosphere (see VS1-BRACKISH and VS1-HIPH in Table 7-3); 2) the composition of porewater in the bentonite, reflected in sorption, diffusion and solubility values for radionuclides in the canister, buffer, and backfill (see VS1-HIPH_NF in Table 7-3). The buffer thickness is varied relative to the base scenario, as its performance is assumed to be degraded because of loss of buffer due to piping and erosion or because of montmorillonite transformation (see process in Features, Events and Processes). Variant scenario 2 (VS2): Corrosion failure following buffer erosion The line of evolution leading to this scenario includes chemical erosion of the buffer followed by enhanced corrosion of the copper canister. It is assumed that all canisters are initially intact, and that there are no processes adversely affecting the safety functions of the EBS and the geosphere until after the advance and retreat of ice sheets has made possible the conditions for penetration of dilute groundwaters to repository depth. During the retreat of ice sheets, the intrusion of dilute glacial meltwater of low ionic strength may reach a few deposition holes for a short period leading to chemical erosion of bentonite (see Sections 7.1 to 7.5 in Performance Assessment). After sufficient buffer material has been eroded, advective conditions may be established between the canister and the rock; this is assumed to occur once the buffer mass loss exceeds 1200 kg for the cautious assumption of incomplete homogenisation of the buffer or fracture clogging (see Section in Performance Assessment). Advective groundwater flow then carries sulphide to the affected canister(s) and, after corrosion has progressed sufficiently, the loss of canister containment and transport resistance. There is uncertainty in the flow conditions around the deposition holes, and hence in the number, locations and timings of canister failures. Four calculations cases are defined, each corresponding to one of the four canisters that are calculated to fail within this time frame based on a single realisation of the DFN groundwater flow model and a specific set of groundwater flow model assumptions, as described in Performance Assessment. Overall radionuclide releases in this scenario are obtained by superimposing the results of these cases Disturbance scenarios for the repository system In formulating disturbance scenarios, two main unlikely events are taken into account: one is the occurrence of an earthquake capable of originating a rock shear large enough to breach the canister, for which the probability of occurrence within different time windows has been documented in Section 7.2 of Performance Assessment; the other is inadvertent human intrusion, which is treated as a surface environment scenario (see Section 7.3.3). FEPs that are likely to occur, but only detrimentally affect safety

213 168 functions if their rates are outside the expected range of possibilities, are also taken into account. Further, the unlikely event of an accelerated insert corrosion rate, which allows the insert to corrode faster and the corrosion products of the insert to expand and mechanically breach the copper overpack resulting in a total loss of transport resistance, has been considered. FEP number reactivation displacement along existing fractures is closely related to an earthquake-promoted rock shear, as rock shear is most likely to occur on existing fractures (see in Features, Events and Processes). This FEP may be combined with other FEPs in defining disturbance scenarios (see below). Two other FEPs that are likely to occur, but only detrimentally affect safety functions if acting at a rate outside the expected range of possibilities, are corrosion coupled to mechanical deformation (see processes 4.2.5/4.2.6 and in Features, Events and Processes). Disturbance scenario: Rock shear (RS) The line of evolution of the repository system selected for this scenario is that of expected evolution up to either 1) about 40,000 years AP or 2) about 155,000 years AP. At this time, it is assumed that an earthquake occurs that causes a rock shear displacement sufficient to breach a canister, but still keeping most of the buffer in place. The selection of times of 40,000 and 155,000 years is based on the annual probability for canister failure due to rock shear, being largest between 10,000 and 50,000 years AP and after 100,000 years AP. The time 40,000 years is selected arbitrarily for comparison to the results of a latter time. The significance of the latter time arises from the retreat of an ice sheet and re-establishment of temperate conditions (see Ch. 4 in this Synthesis and Section in Performance Assessment), i.e. it relates to post-glacial fault reactivation. The major difference between assuming an earthquake at 40,000 or at 155,000 years is the radionuclide inventory of the canister at that time, as the annual probability of occurrence of canister failure is the same in both cases (see Section 7.2 in Performance Assessment). Other uncertainties to be accounted for are groundwater flow and composition, which may be substantially changed in the case of an earthquake after ice-sheet retreat. Disturbance scenario: Rock shear and buffer erosion (RS-DIL) The line of evolution of the repository system selected for this scenario is, in part, identical to the Rock shear scenario, i.e. normal or expected evolution up to either 1) about 40,000 AP or 2) about 155,000 year AP. At this time, it is assumed that an earthquake occurs that causes a rock shear displacement sufficient to breach a canister. In addition, however, a perturbation of the fracture network due to the rock shear is assumed, leading to inflow of dilute, low ionic strength, water reaching the positions of the breached canisters at the time of canister failure and also later in association with ice-sheet retreat and hence resulting in chemical erosion of the buffer. The penetration of dilute water due to the long persistence of temperate conditions with meteoric water infiltration to repository depth at 40,000 year AP is likely, but not necessarily at any particular canister position (see Section in Performance Assessment), and therefore also not necessarily at the canister positions affected by a rock shear displacement. There is a cautious assumption of incomplete homogenisation of the buffer and either to fracture clogging. Advective conditions are assumed to be established in the buffer once the buffer mass loss exceeds 1200 kg (see above).

214 169 Disturbance scenario: Accelerated insert corrosion (AIC) In this scenario, it is assumed that a canister with an initial penetrating defect (1 mm diameter) has been emplaced in the repository, as in the base scenario. The base scenario assumes a reasonable expected corrosion rate for the cast iron insert (0.1 to 1 μm/year; Pastina & Hellä 2010) and a reasonable behaviour for the evolution of the corrosion products (mostly magnetite) with no consequences for the copper overpack. In the AIC scenario, however, an accelerated insert corrosion rate (> 1 μm/year) is assumed, allowing the insert to corrode faster and the corrosion products of the insert to expand and mechanically breach the copper overpack resulting in a total loss of transport resistance after 15,000 years. There is uncertainty over the tightness of the cast iron insert. If the insert is water-tight then radionuclide releases cannot occur until the insert is breached (at 15,000 years); if the insert is not tight (leaky) then releases may begin at 1000 years as in the Reference Case, with total loss of transport resistance after 15,000 years. The alternatives are represented by two calculation cases. The assumptions of this scenario are very pessimistic. First, the minimum thickness of the insert outer wall varies between 30 and 50 mm and thus it would take a minimum of 30,000 to 50,000 years for the insert to corrode at the corrosion rate of 1 μm/year. Moreover, there is no evidence that the formation of the insert corrosion products could breach the copper overpack in repository conditions since magnetite is porous and yielding oxide. A third factor is that the corrosion of the iron insert would be limited by the low rate of diffusion of water vapour through the defect. For these reasons the scenario is judged very unlikely, and classified as a disturbance scenario Radionuclide release scenarios and cases The radionuclide release scenarios and cases taken forward to quantitative analysis are summarised in Tables 7-2 to 7-4. These present a systematic investigation of the main uncertainties identified within each scenario. Table 7-2. Calculation cases for the radionuclide release base scenario for the repository system. Scenario Calculation case Short description BASE SCENARIO: Incidental deviation of introducing one or a few canisters with a penetrating defect of 1 mm diameter BS-RC BS-LOC1 BS-LOC2 BS-ANNFF BS-TIME Reference case (RC) one canister with an initial penetrating defect of 1 mm diameter. Cautious position selected from a DFN realisation taking into account the whole repository. As RC, except alternative position_1 investigates the uncertainties in the selection of flow-related parameters (uncertainty in DFN realisation). As RC, except alternative position_2 investigates the uncertainties in the selection of flow-related parameters (uncertainty in DFN realisation). As RC, except Ag, Mo, Nb migrate as anions in the near and far field (i.e. geosphere) investigates uncertainty in the speciation of those elements. As RC, except uncertainty in the time needed to establish a transport path from the defective canister is taken into account (1000 years in RC and 5000 in TIME)

215 170 Table 7-3. Calculation cases for the radionuclide release variant scenarios. Scenario Calculation case Short description VARIANT SCENARIO 1: Initial defect gradually enlarges due to corrosion VARIANT SCENARIO 2: No initial penetrating defects (thin copper canister wall 35 mm): Erosion of buffer and subsequent corrosion of four canisters VS1-BRACKISH VS1-HIPH VS1-HIPH_NF VS2-H1 VS2-H2 VS2-H3 VS2-H4 Cautious position as in the RC initial penetrating defect enlarging; degraded buffer; speciation for brackish water. Cautious position as in the RC initial penetrating defect enlarging; degraded buffer; speciation for high ph water in the near and far field. Cautious position as in the RC initial penetrating defect enlarging; degraded buffer; speciation for high ph water in the near field alone. Canisters in four positions fail due to corrosion after buffer is chemically eroded. The four canisters that are calculated to fail within this time frame are based on a single realisation of the DFN groundwater flow model and a specific (reference) set of groundwater flow model assumptions. Table 7-4. Calculation cases for the radionuclide release disturbance scenarios Scenario Calculation case Short description AIC Accelerated Insert Corrosion RS Rock Shear RS-DIL Rock Shear followed of buffer erosion AIC-LI AIC-TI RS1 RS2 RS1-DIL RS2-DIL The insert of a defective canister with an initial defect starts to corrode at 1000 years after emplacement releases from a leaky insert start also at 1000 years. Transport resistance is suddenly lost at 15,000 years. The insert of a defective canister with an initial defect starts to corrode at 1000 years after emplacement no releases from a tight insert. Transport resistance is suddenly lost at 15,000 years. Canister(s) fail as a consequence of rock-shear at 40,000 years after emplacement. Canister(s) fail as a consequence of rock-shear at 155,000 years after emplacement. Canister(s) fail as a consequence of rock-shear at 40,000 years. Buffer erosion follows the event whenever low ionic strength water is available. Canister(s) fail as a consequence of rock-shear at 155,000 years. Buffer erosion follows the event whenever low ionic strength water is available. 7.3 Surface environment scenarios This section gives an overview of the main assumptions used in formulating the radionuclide release scenarios for the surface environment; these are discussed in detail in the Formulation of Radionuclide Release Scenarios Base scenario for the surface environment The surface environment scenarios are formulated independently from the repository system and are limited to the dose assessment time window, hence covering the first ten millennia after disposal. The base scenario for the surface environment is formulated bearing in mind that this time window is relatively short compared with the whole

216 171 assessment time frame. The base scenario and its main assumptions are briefly summarised below. Key statements in the regulations are that the environmental changes due to sea-level changes relative to the land (i.e. allowing for land uplift) should be considered, and that the climate type as well as the human habits can be assumed to remain unchanged (Guide YVL D.5, 307). Thus it is appropriate to assume the current climate type in the region of the Olkiluoto site in the scenario formulation. Furthermore, Posiva judges that it is appropriate to assume present-day demographic data and human habits, such as the number of inhabitants in the region and land use, in the scenario formulation. The past development of the climate and surface environment conditions, focusing on the time span since the last deglaciation until present and current conditions in the region of the Olkiluoto site are presented in Biosphere Description, along with the current activities and habits of people in the region of the Olkiluoto site. This forms the knowledge basis, especially on land uses and exposure pathways to people. Furthermore, the regulations state (Guide YVL D.5, 317) that the assessment of radiation exposures of flora and fauna shall assume present kinds of living terrestrial and aquatic populations in the disposal site environment. Posiva interprets this as making it appropriate to identify a set of representative species based on present-day conditions at the Olkiluoto site and in the region. This is discussed in detail in Dose Assessment for Plants and Animals. The main features, events and processes (FEPs) assumed to drive the scenario formulation (the key scenario drivers) are associated with the evolution of the natural environment, the climate and how humans behave, especially how the land is utilised. The two key scenario drivers identified are sea-level change (local) and land use. The key scenario driver sea-level change (local) is an aggregated FEP. The FEPs identified to have the strongest influence on sea-level change at the Olkiluoto site are climate evolution (Features, Events and Processes, Section ) and land uplift and depression (Features, Events and Processes, Section ). The line of evolution selected for the base scenario for the key scenario driver sea-level change (local) assumes that the current climate prevails. The key scenario driver land use is also an aggregated FEP. The FEPs regarding land use at the Olkiluoto site identified to be most important in scenario formulation are crop type, irrigation procedures and livestock (addressed in Features, Events and Processes, Section 9.2.4), forest and peatland management (Features, Events and Processes, Section 9.2.5), construction of a well (Features, Events and Processes, Section ) and demographics (Features, Events and Processes, Section ). The line of evolution selected for the base scenario for the scenario key driver land use is based on present-day conditions and prevailing land use practices. The FEPs mentioned above are those with identified alternative lines of evolution for the key scenario drivers that are primarily addressed when formulating variant and disturbance scenarios. More FEPs (see Table 4-1 in Section 4) are taken into account in the detailed formulation of the scenarios, and subsequently when defining the calculation cases to be used in analysing the scenarios.

217 Variant scenarios for the surface environment Variant scenarios for the surface environment are based on alternative credible lines of evolution arising from reasonable variations of the FEPs affecting the key scenario drivers. Consideration has also been given to additional scenario drivers. The variant scenarios that are analysed in terms of doses are listed in Table 7-5 and briefly described below. It should be noted that more scenarios are formulated and analysed in the TESM, but not propagated further in the biosphere assemment modelling (these are documented in Biosphere Assessment). VS(A) Discharge locations to the surface environment Scenario driver: Discharge locations The discharge locations of radionuclide release from a canister with an initial penetrating defect may be affected by the position of the defective canister in the repository, which affects the groundwater flow path(s) by which contaminants will be carried to the surface environment. The implication of the uncertainty in the location of a canister and groundwater flow paths on the radiological impact is addressed in this scenario. VS(D) Land use (well) Scenario driver: Land use This variant scenario addresses uncertainties in how humans use the land, focusing on the construction of wells to extract drinking water. In the Base Scenario, the number of wells used in the model domain is consistent with the present-day average well density in southwestern Finland. The implication of the uncertainty in the number of wells on the radiological impact is addressed in this scenario. Table 7-5. Variant scenarios identified for the surface environment, limited to the ones analysed in terms of doses, the driver that the scenarios address and the most important FEPs which uncertainties affect the drivers. Variant scenario Scenario driver FEPs VS-A Discharge locations to the surface environment Discharge locations Defective canister location in the repository layout (a) VS-D Land use (well) Land use Construction of a well (9.2.29), Well VS-E Route of radionuclide transport Element migration and accumulation (9.2.30) Alternative radionuclide transport routes in biosphere terrestrial and aquatic compartments affect a number of terrestrial and aquatic processes VS-F Exposure characteristics Human habits Dietary profile (9.2.32) VS-G Combined scenario Sea-level change (local) Land use (a) Agriculture and aquaculture (9.2.4), Climate evolution (10.2.1), Land uplift and depression (10.2.4) See BS-LOC1 and BS-LOC2 in Assessment of Radionuclide Release Scenarios for the Repository System and Table 7-10.

218 173 VS(E) Alternative radionuclide transport routes in biosphere compartments Scenario driver: Element migration and accumulation This variant scenario addresses uncertainties in the assumptions underlying the radionuclide transport in the surface environment. In the Base Scenario it is assumed that the radionuclide releases from the geosphere enter the biosphere through a deep overburden in a terrestrial and agricultural ecosystem or into deep sediment in aquatic ecosystems. In this variant scenario alternative compartments receiving the initial releases are assumed (e.g. the radionuclides are released direct to the rooting zone in terrestrial and agricultural ecosystems). VS(F) Exposure characteristics Scenario driver: Human habits This variant scenario addresses uncertainties in the human diet. In the Base Scenario, it is assumed that all (contaminated) edibles possibly produced from the different ecosystems at the site are consumed by humans, and that humans have no preferences regarding the mix of foods consumed. This scenario assumes that humans in future generations have the same preferences (dietary profile) regarding food consumption as the present-day Finnish population. VS(G) Combined scenario Scenario drivers: Sea-level change (local), Land use This variant scenario addresses uncertainties in climate evolution, land uplift and depression, agriculture and aquaculture. For each scenario driver, and the FEPs affecting it, assumptions are made to individually maximise the arable land area, while remaining consistent with current scientific understanding and within the reasonably expected range of possibilities Disturbance scenarios for the surface environment In the disturbance scenarios for the surface environment, unlikely lines of evolution that may have a significant effect on the fate of radionuclides in the surface environment and/or the radiation exposure of humans, plants and animals are addressed. The identified disturbance scenarios that are analysed in terms of doses are listed in Table 7-6 and briefly described below. It should be noted that more disturbance scenarios are formulated and analysed in the TESM, but not propagated further in the biosphere assemment modelling (these are documented in Biosphere Assessment). DS(D) Exposure characteristics Scenario driver: Biotope occupancy This disturbance scenario assesses the impact on the doses to plants and animals due to uncertainties in the biotope occupancy. In the Base Scenario it is assumed that plants and animals have specific occupancy preferences (e.g. that some aquatic speices are found in freshwater but not brackish water), but that they may be found in any suitable part of the contaminated area of the model domain. This scenario cautiously assumes constant occupancy of plants and animals in the most constraining (in terms of dose rate) biotope.

219 174 Table 7-6. Disturbance scenarios identified for the surface environment, the driver the scenarios address and the most important FEPs which uncertainties affect the drivers. Disturbance scenario Scenario driver FEPs DS(D) Exposure characteristics Biotope occupancy Habitats DS(F) Inadvertent Human Human actions Human actions intrusion DS(G) Deep wells Land use Constructing a well DS(F) Inadvertent Human intrusion Scenario driver: Human actions Human intrusion has been considered an issue in post-closure safety of solid radioactive waste disposal for many years (NEA 1989). It has been concluded that the possibility of human intrusion should not be ignored, but it is necessary to recognise the illustrative nature of any assessment. ICRP (2000, paragraph 62) recommends that one or more typical plausible stylised {human intrusion} scenarios should be considered to evaluate the resilience of a repository to possible human intrusion events. In TURVA- 2012, the reference approach for evaluation of human intruder doses developed wihin the BIOPROTA forum has been adopted (Smith et al. 2012). This includes the assessment of scenarios for people having direct contact with contaminated material brought to the surface by drilling, and others who might be exposed due to contaminated material being left at the drill site. DS(G) Deep wells Scenario driver: Land use Wells are constructed in the bedrock (drilling or driving or dug) in the overburden, which may be used for extracting household water, watering animals and irrigation purposes. Here, the unlikely event that a deep (> 300 m) well is drilled at, or in the vicinity of, the site is addressed. It is assumed that the well intersects a waterconducting feature somewhere deep in the geosphere and the water drawn from the well has passed through the repository Calculation cases As stated above it is not feasible to identify the most likely lines of evolution for the entire surface environment, thus it is not possible to rank the base and variant scenarios according to their likelihood of occurrence. In order to facilitate a clear communication of the safety assessment results, the biosphere Reference Case (BSA-RC) is the only one used to interpret the Base Scenario for the surface environment, and the other biosphere calculation cases are identified as sensitivity cases arising under variant scenarios, and what-if cases under disturbance scenarios. The analysis of the biosphere calculation cases is done step-wise in a modelling chain taking into account the connection between each biosphere assessment sub-process (see Figure 5-8). 21 cases are first defined and calculated within the terrain and ecosystem development modelling, which result is a series of projections of the development of the surface environment (see Terrain and Ecosystem Development Modelling). It must be noted that from the 21 cases that have been calculated (see Biosphere Assessment), only

220 175 the cases presented in Table 7-7 have been propagated to sub-sequent models in the biosphere assessment process (the reason for not propagating all 21 cases is that most projections are considered to be similar to the Reference Case projections or bounded by the Terr_MaxAgri calculation case). Biosphere calculation cases, including the rest of the chain in the biosphere modelling process, are then identified for each scenario (Tables 7-8 and 7-9). In each sub-process the FEPs to be adressed given by the scenario description are identified and models and parameter values are selected accordingly. This is done independently for the surface- and near-surface hydrological model (SHYD), the landscape model (LSM), and the dose models used in the radiological impact assessment (RIA). It is also checked that the settings for each sub-process model are consistent with both each other and with the scenario to be analysed. The source term or input for the analysis of the biosphere calculation cases is the result of the repository calculation cases that give radionuclide releases within the dose assessment time window (i.e. up to 10,000 years after disposal). Most of the repository calculation cases propagated to biosphere assessment are analysed with the biosphere Reference Case (BSA-RC) (see Table 7-10). The outcome of the Reference Case in the TESM is utilised to construct two landscape models: one model for geosphere releases from the discharge locations north of the present Olkiluoto Island, hence suitable for analysing the geosphere releases from BS- RC, and all other repository calculation cases that are based on the same canister location, and one model for geosphere releases from the alternative discharge locations south of the present Olkiluoto Island (the geosphere releases in the cases BS-LOC1 and BS-LOC2). These two landscape models are denoted REF and SOUTH in Table 7-8. The biosphere objects receiving the direct releases differ in the cases BS-LOC1 and BS- LOC2, which have implications for the radionuclide transport modelling part of the landscape modelling sub-process. This is reflected by two model variants for the model SOUTH (denoted Southern_1 and Southern_2 in the RNT column in Table 7-8). A Reference Case model is set-up in the surface and near-surface hydrology (SHYD), denoted REF in Table 7-8. In addition to this, two SHYD cases are identified to analyse the variant scenario VS(D) addressing undertaintiesd in the number of wells at the site (denoted MORE_WELLS and NO_WELLS in Table 7-8). Table 7-7. Calculation cases assessed in terrain and ecosystem modelling (TESM). TESM case FEPs taken into account Reference Case Terr_maxAgri All relevant FEPs are taken in to account as specified in TESM. Sea level, Land uplift, River boundary, Aquatic erosion and Land use differ from the Reference Case Propagated to SHYD YES YES Comments Combines maximised terrestrial area within reasonably expected development (faster land uplift, enhanched sedimentation in lakes and maximum extent of agricultural land. Uses climate simulation A2 (see Ch.4 in Biosphere Assessment)

221 176 In the terrain and ecosystem development line that maximises the land area for agricultural practices (Terr_maxAgri; Table 7-7), alternative sub-models and data are used for land uplift, climate specifics, sedimentation in lakes and the extent of agricultural land. Also the water body connections, the surface and near-surface hydrology and the landscape model change accordingly compared to the base scenario. The only models that are not affected by this alternative TESM projection are the dose models in the RIA. In the base scenario the geosphere releases are assumed enter the biosphere in the deepest layer of the overburden. To account for uncertainties in the configuration of the overburden and the release path within it that may affect radionuclide migration and accumulation, one case, RNT1, has been defined for the RNT part assuming that the geosphere releases enter the biosphere directly in the rooting zone for terrestrial and agricultural ecosystems, and directly to the water column in aquatic ecosystems. In the radiological impact assessment (RIA) the reference model is used for most of the biosphere calculation cases. Uncertainties in the dietary habits for humans are taken into account by implementing a model assuming that future generations have the same preferences regarding consumption of various food groups as the present-day Finnish population. Calculation cases within the base scenario The Reference Case (BSA-RC) is the only one within the base scenario. This, and the repository calculation reference case (BS-RC) connect the entire disposal system in what is deemed the most expected and credible line of evolution. Calculation cases within variant scenarios Within the variant scenario VS(A), two calculation cases are defined that account for uncertainties in the discharge locations to the surface environment (see Tables 7-8 and 7-10). In both cases, VS(A)-SOUTH1 and VS(A)-SOUTH2 the reference terrain and ecosystem development modelling results are used, but uses a different landscape model compared to the Reference Case, since the contaminated areas are located south of the present Olkiluoto Island. The assumptions and models the RIA are the same as in the Reference Case. Two cases have been defined for the variant scenario VS(D), VS(D)-WELL that assumes there to be more wells than in the base scenario, and VS(D)-NO_WELL that assumes there are no wells (see Table 7-8). In variant scenario VS(E), an alternative radionuclide transport route in biosphere compartments are considered in VS(E)-RNT1. In this case radionuclide releases from the geosphere to the biosphere is assumed to be introduced directly to the rooting zone in terrestrial and agricultural ecosystems and to the water column in aquatic ecosystems. Within the variant scenario VS(F), two cases are defined that differ from the Reference Case only in the dietary profile assumed for future human generations. VS(F)-FINDIET assumes that the consumed various food groups are the same as consumed by the present-day Finnish population (based on the The National FINDIET 2007 Survey,

222 177 Paturi et al. 2008), and VS(F)-VEG assumes a vegetarian diet. No doses to animals and plants will be calculated in these cases as they would remain identical to the reference case. VS(G)-COMBI is the only case defined within the variant scenario VS(G), where the uncertainties in climate evolution, land uplift and depression, agriculture and aquaculture are combined. In this case the extension of the areas dedicated to agriculture is maximised. The terrain and ecosystem modelling (TESM), the results used are those of Terr_maxAgri and so are the results of SHYD and LSM (Table 7-8). Table 7-8. Biosphere calculation cases under the base and variant scenarios. Calculation case Calculation case in the sub-process modelling Comments TESM SHYD LSM RNT RIA BSA-RC REF REF REF REF REF Reference Case for the Base scenario for the entire disposal system. VS(A)-SOUTH1 REF REF SOUTH Southern_1 REF Uncertainties in discharge locations to the biosphere VS(A)-SOUTH2 REF REF SOUTH Southern_2 REF Uncertainties in discharge locations to the biosphere VS(D)-WELL REF MORE_WELLS REF REF REF More wells than in the Reference Case VS(D)-NO_WELL REF NO_WELLS REF REF REF No wells. Irrigation and drinking water from surface waters. VS(E)-RNT1 REF REF REF REF1 REF Geosphere releases get directly to alternative compartments compared with the BSA-RC. See main text. VS(F)-FINDIET REF REF REF REF FINDIET Dietary profile based on present-day average consumption statistics VS(F)-VEG REF REF REF REF VEG Dietary profile with no meat and fish. VS(G)-COMBI Terr_max Agri Terr_maxAgri Terr_max Agri Terr_maxAgri REF Combination of uncertainties in sea level, climate and assumes extensive agriculture

223 178 Calculation cases within disturbance scenarios The disturbance scenario DS(D) takes into account unlikely exposure characteristics for plants and animals, which depend on their habitat. One calculation case DS(D)- HABITAT has been defined that considers that the most exposed plants and animals live in the most contaminated object. In this case the same reference TESM, SHYD, and LSM as in the reference case are used (see Table 7-9a). In DS(D)-HABITAT the radiological impact assessment for humans remains as in the reference case. In the inadvertent human intrusion scenario DS(F) six calculations cases are considered. In all cases it is assumed that borehole drilling (e.g. for geothermal energy purposes), is conducted somewhere within the footprint of the repository and reach repository depth. In two of the six cases it is assumed that drilling hits an intact canister (DS(F)- CANISTER) containing all spent fuel radionuclide inventory at the time. In one case all the drill crew is exposed (DS(F)-CANISTER-D) and in the other a geologist is exposed when examining drill core(s). In the other cases it is assumed that contaminated buffer and backfill material is drilled and brought up to the surface (see DS(F)-BUFFER and DS(F)-BACKFILL in Table 7-9b). Table 7-9a. Biosphere calculation case under disturbance scenarios that is analysed following all biosphere assessment sub-processes. Calculation cases Calculation case in the sub-process modelling TESM SHYD LSM RNT RIA Comments DS(D)-HABITAT REF REF REF REF Occupancy constraints Only dose for plants and animals Table 7-9b. Biosphere calculation cases for analysing inadvertent human intrusion. Calculation cases Comments DS(F)-HI-CANISTER-D DS(F)-HI-CANISTER-G DS(F)-HI-BUFFER-D Drilling through an intact spent fuel canister; Drill crew exposed in the process Drilling through an intact spent fuel canister; Geologist exposed when examining a drill core Drilling through contaminated buffer material; Drill crew exposed in the process DS(F)-HI-BUFFER-G Drilling through contaminated buffer material; Geologist exposed when examining a drill core DS(F)-HI-BACKFILL-D Drilling through contaminated backfill material; Drill crew exposed in the process DS(F)-HI-BACKFILL-G Drilling through contaminated backfill material; Geologist exposed when examining a drill core

224 179 In the deep well scenario DS(G) all repository repository calculation cases that give releases within the dose assessment time frame window are analysed as separate calculation cases. These are not further addressed in this report, the analysis of these cases are presented in detail in Biosphere Assessment to which the interested reader is encouraged to consult. Link between repository and biosphere calculation cases There are nine repository calculation cases that give releases within the dose assessment time window of 10,000 years. These cases are propagated to the biosphere assessment and analysed with biosphere calculation cases. These repository calculation cases and the type of biosphere calculation cases that are used to analyse them are summarised in Table The guiding principle is that the Reference Case for the repository system (BS-RC) is analysed with all identified biosphere calculation cases, and all other repository calculation cases are analysed with the Reference Case for the surface environment (BSA-RC). An exception is that some repository calculation cases lead to geosphere release to the south of the present-day Olkiluoto Island, so that is more suitable to use the characteristics of this area in the biosphere analysis. Table Repository calculation cases propagated to the biosphere assessment. Biosphere calculation cases used in the analysis and the name of the resulting calculation case that is analysed with the biosphere full modelling chain. Repository calculation case Description of Repository Calculation Case BS-RC Canister location 381, leading to releases to the surface environment north of the present-day Olkiluoto Island BS-LOC1 Sensitivity case assuming alternative canister location. Canister location 2418, leading to releases to the surface environment south of the present-day Olkiluoto Island BS-LOC2 Sensitivity case assuming alternative canister location. Canister location 3829, leading to releases to the surface environment south of the present-day Olkiluoto Island BS-ANNFF Sensitivity case assuming alternative near-field and geosphere speciation. BS-TIME Sensitivity case assuming delayed establishment of transport path. VS1- BRACKISH Sensitivity case assuming reduced buffer thickness. Biosphere Calculation Case/s used in the analysis (see also Table 7-8 ) All identified biosphere cases for the relevant discharge locations and human intrusion scenario cases VS(A)-SOUTH1 VS(A)-SOUTH2 BSA-RC BSA-RC BSA-RC Name of the resulting calculation case combination BSA-RC (see Table 7-8) VS(A)-SOUTH1 (see Table 7-8) VS(A)-SOUTH2 (see Table 7-8) BSA-ANNFF BSA-TIME BSA-BRACKISH

225 180 VS1-HIPH Sensitivity case assuming reduced buffer thickness and high-ph groundwater. Sensitivity case assuming reduced buffer thickness and high-ph (near-field). What-if case assuming a high insert corrosion rate, causing a sudden loss of transport resistance of the defect after 15,000 years. Identical to BS-RC in the dose assessment time window. BSA-RC BSA-HIPH VS1-HIPH_NF BSA-RC BSA-HIPH_NF AIC-LI BSA-RC BSA-AIC-LI 7.4 Summary and discussion on comprehensiveness The formulation of radionuclide release scenarios for the repository system is the link between Performance Assessment and Analysis of Radionuclide Release Scenarios for the Repository System. All these make use of the Features, Events and Processes (FEPs) that potentially could affect the disposal system and that have been defined in Features, Events and Processes. The formulation of radionuclide release scenarios for the surface environment brings together Biosphere Description and the surface environment FEPs and is the link to the assessment of the surface environment scenarios. The assessment of the surface environment scenarios is summarised in Biosphere Assessment and discussed in detail in: Terrain and Ecosystems Development Modelling; Surface and Near-Surface Hydrological Modelling; Biosphere Radionuclide Transport and Dose Assessment; and Dose Assessment for Plants and Animals Demonstrating that the set of scenarios is comprehensive To claim that the set of formulated repository system scenarios is comprehensive it is necessary to check if all the relevant FEPs have been taken into account. The scenarios are formulated to ascertain the impact of uncertainties in the initial state and evolution of the repository system, which have been highlighted in Performance Assessment. In that report the most relevant evolution-related FEPs were taken into account, therefore it remains to be checked what other FEPs have been included in the base, variant, and disturbance scenarios. Appendix 3 shows the repository system related FEPs with colour codes. Most of the non-coloured FEPs have been taken into account in groundwater flow modelling (e.g. degradation of auxiliary components, most backfill evolution-related processes, erosion and sedimentation in fractures, etc.). Possible scenarios dealing with criticality will be treated in a future stage after submitting the construction licence application to confirm

226 181 that, in the long term, the possibility of criticality within or outside waste packages is indeed negligible. Freezing and thawing has been shown not to have detrimental effects for the buffer, backfill, and auxiliary components. In all the scenarios, the alteration and dissolution of the spent nuclear fuel matrix (FEP in Features, Events and Processes), radioactive decay (FEP 3.2.1), and the release of the labile fraction of the inventory (FEP 3.2.9) are taken into account. Heat generation (FEP 3.2.2) and heat transfer (3.2.3) from the spent fuel are taken into account in the design of the layout of the disposal facility as well as in the modelling of groundwater flow (see FEPs in yellow in Appendix 3). Although precipitation and coprecipitation are conservatively not taken into account in radionuclide migration calculations, they are nonetheless accounted for in calculating solubility limits. These limits are applied in all the scenarios. In the base scenario all migration FEPs are taken into account with the exception of colloid transport (FEPs 4.3.6/5.3.6/6.3.6/8.3.6 in Features, Events and Processes report) and advection in the canister, buffer and backfill (FEPs 4.3.5/5.3.5/6.3.5). In variant scenario VS1, the partial degradation of the buffer is taken explicitly into account due either to piping and erosion (FEP 5.2.3) or to montmorillonite transformation (5.2.3). The influence of microbial activity (5.2.8/8.2.10) is taken into account indirectly as a potential cause of higher than expected corrosion rates. All migration FEPs are also taken into account, again with the exception of colloid transport (see FEPs in yellow and green in Appendix 3). In variant scenario VS2, the buffer is gradually degraded due to chemical erosion (FEP 5.2.4), and transported as colloids (5.3.6) because of adverse groundwater chemistry (FEP 8.2.7). The canister is gradually corroded (FEPs 4.2.5, 4.2.6, 4.2.7) with or without the aid of microbial activity (8.2.10) (see FEPs in yellow, green and blue in Appendix 3). The rock shear in the disturbance scenarios RS and RS-DIL take into account the reactivation and displacement of fractures (FEP 8.2.3) as well as deformation of the canister(s) (FEP 4.2.3) as it breaches (see also FEPs in orange and yellow). The canister may fail easily as a consequence of rock shear if it has been affected by corrosion, which is not explicitly taken into account, but the VS2 scenario encompasses this possibility. In RS-DIL, the FEPs in blue and advection in the buffer (FEP in green) are also taken into account. The disturbance scenario AIC takes into account corrosion and deformation of the canister. The accelerated insert corrosion could also be explained by microbial activity (see also FEPs in yellow in Appendix 3). The surface environment scenarios take into account the FEPs listed in Chapter 9 in Features, Events and Processes to which the calculation cases in Sections to of this report refer. Based on the explanations given above, it can be said that the repository system scenarios are comprehensive, since all the FEPs influencing long-term safety have been

227 182 taken into account. To be confident that the most penalising conditions have been assessed, the combinations of repository system scenarios and their implied FEPs are considered in the next section. The identified set of surface environment scenarios is not yet as thoroughly scrutinised for comprehensiveness as the repository system scenarios are. This work is in progress and will mature during the next iteration of the safety case. However, it is the view of Posiva that the set of scenarios is sufficiently comprehensive to support an application for construction Combinations of repository system scenarios The radionuclide release scenarios discussed above have each been considered individually. The scenarios are not, however, all mutually exclusive. For example, it is not impossible that one of the earthquake/rock shear scenarios occurs when the repository also contains one or more canisters with initial penetrating defects. If the rock shear event affected the defective canisters, then the radionuclide release due to rock shear would be reduced compared with a scenario of rock shear alone, since the IRF radionuclides would already have been released. The likelihood of a rock shear event affecting a defective canister is, however, much lower than that of it affecting an intact canister, since there will be very few if any defective canisters present. Thus, it is appropriate to treat them as independent, so that their impact in combination is the sum the time-histories of releases from the two scenarios. All possible binary combinations of the base, variant and disturbance scenarios have been considered qualitatively (see Table 7-11). It is found, however, that many of the combinations can be excluded from further analysis, for example, because the releases due to one scenario are far larger than those due to another. Thus, only three combinations are carried forward to analysis. This process is summarised and discussed in detail in Assessment of Radionuclide Release Scenarios for the Repository System.

228 183 Table Possible binary combinations of radionuclide release scenarios. Combinations shown in red are excluded from further analysis for the reasons summarised in the numbered notes, below. The results of the analysis of the combinations shown are presented in Chapter 8. Scenario Base scenario (BS) Enlarging defect/degradation of buffer (VS1) Corrosion failure following buffer erosion (VS2) Accelerated insert corrosion rate (AIC) Earthquake and rock shear (RS) Rock shear followed by buffer erosion (RS-DIL) 1 Retained for analysis 1 3 Retained for analysis Retained for analysis 3 Retained for analysis 2, Retained for analysis Retained for analysis 1 BS VS1 VS2 AIC RS RS-DIL Notes: 1: Excluded since the two scenarios mutually exclusive or inconsistent (e.g. relates to an uncertain process that is assumed significant in one scenario but not in the other). 2: Excluded since combinations involving RS-DIL rather than RS are more penalising, and both include the same rock shear canister failure mode. 3: Excluded since combinations involving AIC rather than VS1 are more penalising, and both include the same key process of defect enlargement.

229 184

230 185 8 ASSESSMENT OF RADIONUCLIDE RELEASE SCENARIOS This chapter summarises and presents overall conclusions from the assessment of the radionuclide release scenarios and calculation cases as defined in Chapter 7. This includes analyses of radionuclide release and transport in the repository system, projections of the development of the surface environment, and analyses of potential radiological impacts on humans, plants and animals. Base, variant and disturbance scenarios are analysed and uncertainties within those scenarios are investigated by a range of deterministic calculation cases as well as Monte Carlo simulations. Probabilistic sensitivity analyses are carried out to assess sensitivities to parameter values and to explore the consequences of alternative model assumptions. The assessments of radionuclide release scenarios and cases are presented in full in Assessment of Radionuclide Release Scenarios for the Repository System and Biosphere Assessment. 8.1 Analysis of the Reference Case in the base scenario The definition of the Reference Case for the base scenario for the repository system and surface environment has been specified in the previous chapter. The radionuclide release scenarios and cases taken forward to quantitative analysis are listed in Table 7-2. The results of the analysis are presented below Results for the repository system Figure 8-1 shows the calculated evolution of radionuclide release rates from the repository near field to the geosphere in the Reference Case, summed over the three release paths: from the buffer into geosphere fractures intercepting the deposition hole (F-path); from the buffer to the EDZ of the deposition tunnel and thence into the geosphere (DZ-path); from the buffer to the tunnel backfill and thence to the geosphere (TDZ-path). The figure shows the evolution of total release rate, summed over all calculated radionuclides, and the release rates of the five radionuclides that make the largest contributions to the total: C-14, Cl-36, Ni-59, I-129 and Cs-135. The peak release occurs at about 4500 years after closure and is dominated by C-14. At later times, the release of C-14 decreases due to radioactive decay, so that beyond about 60,000 years the total release is dominated, first by Cl-36, and later by Cs-135 and I-129. The peak release rate of Ni-59 is smallest of the five, and occurs at around 100,000 years.

231 186 Figure 8-1. Evolution of the total radionuclide release rate from the repository near field to the geosphere in the reference case of the base scenario, summed over the F-, DZ- and TDZ-paths, and the evolution of release rates of C-14, Cl-36, Ni-59, I-129 and Cs-135, which are the radionuclides that make the largest contributions to the total. Figure 8-2 shows the near-field release and geosphere release rates normalised with respect to the radionuclide-specific constraints for releases to the environment defined in STUK Guide YVL D.5. The figure indicates that during the dose assessment time window (up to 10,000 years) the normalised activity release is almost four orders of magnitude below the criterion of one defined in Para 313 of STUK s Guide YVL D.5; beyond a few tens of thousands of years the normalised activity release rate decreases to between five and six orders of magnitude below one. The limited role of the geosphere in attenuating the peak release rate is related to the cautiously selected location of the deposition hole containing the defective canister. The defective canister is, in reality, equally likely to be emplaced in any of the deposition holes accepted for emplacement. As discussed in Section of the Assessment of Radionuclide Release Scenarios for the Repository System, a defective canister emplaced in the majority of deposition holes would give rise to far lower peak release rates, since geosphere transport would be sufficiently slow for substantial decay of the C-14 to occur during transport.

232 187 Figure 8-2. Evolution of the near-field and geosphere release rates in the reference case of the base scenario (BS-RC), with the release rates for each radionuclide normalised with respect to the regulatory nuclide-specific constraints for radioactive releases to the environment (see Section 2.3.7). For the period beyond the dose criteria time window (taken to be 10,000 years in this and similar figures), the peak normalised activity release rate, summed over all radionuclides, should be less than one in order to satisfy the regulatory constraint on the overall release rate of activity from the geosphere to the environment (Para 313 STUK Guide YVL D.5) and this is termed the regulatory geo-bio flux constraint in this Figure and similar Figures in this chapter. In summary, in the Reference Case, the normalised activity release rate is, at maximum, about four orders of magnitude below one. The highest activity radionuclide release is of C-14, which peaks at around 4500 years and then declines due to radioactive decay. Beyond a few ten thousand years, longer-lived radionuclides Cl-36, I-129 and Cs-135 dominate radionuclide release. The dominant migration path is from the buffer directly into fractures intercepting the deposition hole (the F-path); migration paths in the EDZ of the deposition tunnel or in the tunnel backfill are less important. For the chosen location of the defective canister, the geosphere has a limited role in attenuating the peak release rate. It would have a greater role for the majority of deposition hole locations, and the peak normalised activity release rate would be correspondingly reduced. Since releases occur within the 10,000-year dose assessment time window, the results of the Reference Case are propagated to the biosphere assessment.

233 Results for the surface environment This section summarises the results for the Reference Case for the surface environment. Firstly, the discharge locations to the surface environment are addressed since this is a key input from the geosphere modelling. Discharge locations to the surface environment As discussed above the Reference Case for the repository system analyses a single defective canister in the repository, located in a relatively unfavourable deposition hole (position 381 in Figure 8-3) in respect to the activity flux from the geosphere (Assessment of Radionuclide Release Scenarios for the Repository System, Section 6.2). Figure 8-3 shows the location of the canister in the repository and the corresponding discharge locations to the surface environment via the F-, DZ- and TDZ-paths (see Figure 5-6). Discharge takes place to the sea off the northern coast of the present island. The discharge locations to the surface environment vary significantly between the F-, DZ- and TDZ-paths and as a function of time. However, the discharge locations tend to converge on a relatively limited area at later times (the green points in Figure 8-3). These considerations have led to the decision to adopt the groundwater flow distribution at 5000 AD as a basis for the selection of flow-related transport parameters for nearfield release and transport modelling and for geosphere transport modelling in the Reference Case. Hence, the area covered by the three green points in Figure 8-3 is the area where the radionuclides are introduced into the landscape model. Figure 8-3. Discharge locations to the surface environment in the Reference Case via the F-, DZ- and TDZ-paths, evaluated for groundwater flow conditions at 2000 AD, 3000 AD and 5000 AD. The present day outline of Olkiluoto Island and the layout of the deposition tunnels considered in the safety case are shown in grey.

234 189 The key results in the biosphere modelling are the projections of the development of the surface environment during the first 10,000 years and the potential radiological impacts on humans, plants and animals living in that environment. The results from analysing the Reference Case (BSA-RC) for the surface environment base scenario are presented in Biosphere Assessment, Section 6.2, and summarised below. Surface environment development The projection of the development of the terrain and ecosystems in the surface environment for the Reference Case is presented in detail in Terrain and Ecosystems Development Modelling. Two illustrative examples are shown in Figures 8-4a and 8-4b. Figure 8-4a. Surface environment projections for three time steps in the Reference Case. The approximate location of central part the repository is indicated with a red circle and the discharge locations with a green circle.

235 190 Figure 8-4b. Ecosystem projections in the Reference Case in 3520 and AD.

236 191 Doses to humans The screening analysis performed on the geosphere releases in the repository case BS-RC (Biosphere Assessment, Section 6.1) resulted in 6 of the 11 radionuclides with a non-zero activity released into the surface environment being screened out from further analysis with the landscape modelling and subsequent dose calculations. The five radionuclides propagated all the way through the biosphere modelling chain in BSA-RC are C-14, Cl-36, Mo-93, Ag-108m and I-129. The annual doses to representative persons within the most exposed group (E most_exp ) and among other exposed people (E other ) are presented in Figure 8-5a and 8-5b. The dose maximum for E most_exp is msv and occurs at about year 5000 and the corresponding dose maximum for E other is msv and occurs at about year These results are about 6-7 orders of magnitude below the regulatory radiation dose constraints. As seen in these figures, C-14 dominates the annual doses. This is a direct effect of that C-14 dominates the geosphere releases in the repository calculation case BS-RC during the dose assessment time window. Furthermore, the shapes of the dose curves are more irregular compared with the shape of the release rate curves (see for example the normalised geosphere release rates in Figure 8-6a). These extra structures in the dose curves are mainly the effect of the dynamics in the landscape model and that the dose is calculated by summing exposure pathway-specific contributions from several biosphere objects. The dynamics in the development of biosphere objects, especially changes in their geometries, have a strong influence on the resulting activity concentrations in both environmental media in contaminated biosphere objects and in the foodstuffs the objects produce. For example, when a lake develops into an agricultural or terrestrial ecosystem it may lead to a steep increase in the activity concentration in the shrinking water volume when it dries out. Doses to plants and animals The (typical) absorbed dose rate for plants and animals for the calculation case BSA-RC for the most exposed organisms in freshwater, brackish, semi-aquatic and terrestrial environments are presented in Figure 8-5c. The dose rate maximum over all organsims is mikrogy/h, which is observed for Pike in freshwater environment.

237 192 Annual dose [msv] 1,E-06 1,E-08 1,E-10 1,E-12 1,E-14 BSA RC E_most_exp C-14 I-129 Cl-36 Ag-108m Mo-93 1,E Year Figure 8-5a. The annual dose to a representative person within the most exposed group (E_most_exp) and the contributions from each radionuclide for the calculation case BSA-RC. Annual dose [msv] 1,E-06 1,E-08 1,E-10 1,E-12 1,E-14 1,E-16 BSA RC E_other C-14 I-129 Cl-36 Ag-108m Mo-93 1,E-18 1,E Year Figure 8-5b. The annual dose to a representative person among other exposed people (E_other) and the contributions from each radionuclide for the calculation case BSA-RC.

238 193 Absorbed dose rate (microgy/h) 1.E-06 1.E-08 1.E-10 1.E-12 BSA-RC Pike (freshwater) Grey Seal (brackish water) Beaver (semi-aquatic environment) Hazel grouse egg (terrestrial environment) 1.E Year Figure 8-5c. Absorbed dose rate for plants and animals for the calculation case BSA-RC for the most exposed organisms in freshwater, brackish, semi-aquatic and terrestrial environments. 8.2 Analysis of other cases in the base scenario In the base scenario, four sensitivity cases are defined to take into account: 1) the uncertainty in the location of the defective canister in the repository system and the corresponding uncertainty of the discharge location in the surface environment; 2) the uncertainty in the speciation of silver, molybdenum and niobium radioisotopes; 3) the uncertainty in the time for the establishment of a transport path between the canister interior and exterior. See cases BS-LOC1 and BS-LOC2, BS-ANNFF, and BS-TIME in Table Alternative canister positions BS-LOC1 and BS-LOC2 A canister with an initial penetrating defect is equally likely to be located in any of the repository deposition holes that are accepted for disposal. The deposition holes vary widely in their near-field flows, in the transport resistances of the potential radionuclide migration paths through the host rock and in the discharge locations to the surface environment. A systematic approach has been used to select the Reference Case position for a defective canister and also alternative positions that, like the Reference Case postion, are cautiously chosen with respect to the peak radionuclide release rates to the surface environment, but discharge to different locations in the surface environment (Assessment of Radionuclide Release Scenarios for the Repository System). Repository system model assumptions and parameter values for calculating radionuclide release rates from the alternative canister positions are identical to those of the Reference Case, except for flow-related transport parameters that are specific to the deposition hole in which the defective canister is assumed to be located. The deposition holes chosen as alternatives to the Reference Case location of the defective canister are positions 2418 (BS-LOC1) and 3829 (BS-LOC2). The results for the geosphere releases are presented in Figure 8-6a. The results from biosphere assessment for these cases are presented in Section 8.4, since alternative canister positions lead to alternative discharge locations in

239 194 the surface environment, which are covered by the surface environment variant scenario VS(A) Alternative speciation BS-ANNFF / BSA-ANNFF In the Reference Case, the only radionuclides considered to migrate in anionic form are I-129, Cl-36 and Se-79. These anions migrate without retardation by sorption through accessible pore space in the buffer and backfill, and this pore space is reduced due to anion exclusion. Repository calculation case BS-ANNFF illustrates the impact of uncertainties in the speciation of silver, molybdenum and niobium that could be present in anionic form and thus also migrate in that form. The assumption of migration in anionic form generally increases the maximum release rates of the radioisotopes of these elements because they are treated as non-sorbing, allowing faster migration and thus less time for radioactive decay. This leads to significant increases in releases of Nb-94, Nb-93m and Mo-93, but the releases of these radionuclides are still minor compared with the releases of C-14, Cl-36, I-129 and Cs-135, so that the total radionuclide release rates are almost unchanged. When the geosphere release rate is normalised with respect to the nuclide-specific constraints for radioactive releases, the geosphere release rates in case BS-ANNFF are indistinguishable from those of the Reference Case (Figure 8-6a). The annual doses to representative persons within the most exposed group (E most_exp ) and among other exposed people (E other ) for biosphere calculation case BSA-ANNFF are presented in Figure 8-6b. The dose maximum for E most_exp is msv and the dose maximum for E other is msv. It should be noted that the screening analysis of the geosphere releases of BS-ANNFF results in the screening in of Nb-94, in addition to the same five radionuclides as in the Reference Case. The (typical) absorbed dose rate maxima for plants and animals for the calculation case BSA-ANNFF is mikrogy/h. This is observed for Reindeer Lichen in terrestrial environments Delayed establishment of the transport path BS-TIME / BSA-TIME In the Reference Case, it is assumed to take 1000 years for water to penetrate the canister insert and fuel cladding and to contact the fuel and structural materials, and for a transport pathway to be established between the canister interior and exterior. This is a cautious assumption. In practice, the slow initial water ingress rate, the decrease of ingress rate over time (e.g. due the build-up of internal pressure due to hydrogen gas formation), and the barriers provided by the cast iron insert and the fuel cladding, are expected to cause a longer delay before establishment of a transport path than assumed in the Reference Case. In the repository calculation case BS-TIME, a 5000-year delay is assumed before a transport path is established between the canister interior and exterior, compared with 1000 years in the Reference Case. The 5000-year delay is arbitrarily chosen from within the range of uncertainty for the purpose of propagating this case to the biosphere

240 195 assessment and analysing the radiological impact at that time. All other model assumptions and parameter values are identical to those of the Reference Case. The delay in establishing the transport path pushes back the time of first release, previously at about 4500 years after disposal, to about 10,000 years after disposal. Consequently, there is a decrease in peak normalised release rate, by about a factor of 2, due to decay of C-14. Beyond about 20,000 years, the peak normalised geosphere release rate in the case BS-TIME is indistinguishable from that in the Reference Case (Figure 8-6a). The annual doses to representative persons within the most exposed group (E most_exp ) and among other exposed people (E other ) for biosphere calculation case BSA-TIME are presented in Figure 8-6b. The dose maximum for E most_exp is msv and the dose maximum for E other is msv. The screening analysis of the geosphere releases of these cases results in only four radionuclides being screened in: C-14, Cl-36, Mo-93 and I-129. The (typical) absorbed dose rate maxima for plants and animals for the calculation case BSA-TIME is mikrogy/h. This is observed for Pike in freshwater environments. Figure 8-6a. Evolution of the geosphere release rates in the Reference Case (BS-RC) and in all the base scenario sensitivity cases, with the release rates for each radionuclide normalised with respect to the regulatory nuclide-specific constraints for radioactive releases to the environment.

241 196 Annual dose [msv] 1.E-04 1.E-06 1.E-08 1.E-10 E most_exp BSA-RC BSA-TIME BSA-ANNFF 1.E Year Annual dose [msv] 1.E-04 1.E-06 1.E-08 1.E-10 E other BSA-RC BSA-TIME BSA-ANNFF 1.E Year Figure 8-6b. The annual doses to representative persons within the most exposed group (E most_exp ) (top) and among other exposed people (E other ) (below) for the biosphere Reference Case (BSA-RC), BSA-ANNFF and BSA-TIME (i.e. the three repository calculation cases in the Base scenario that have the same discharge locations). 8.3 Analysis of the variant scenarios in the repository system Two variant scenarios are identified that are considered plausible: an enlarging defect and degradation of the buffer (VS1) and canister failure by corrosion following buffer erosion (VS2). Sensitivity cases within VS1 investigate alternative assumptions for porewater chemistry in the near field and geosphere. Sensitivity cases within VS2 investigate alternative locations for the failed canister Cases in Variant Scenario 1 (VS1) Three calculation cases are analysed within this scenario, VS1-BRACKISH, VS1- HIPH-NF and VS1-HIPH, which consider the influence of groundwater composition (brackish or highly alkaline) on radionuclide releases.

242 197 Figure 8-7 shows the evolution of near-field release rates via the F-path, the DZ-path and the TDZ-path in the Reference Case and in case VS1-BRACKISH. The influence of groundwater composition (brackish or highly alkaline) on radionuclide releases is shown to be relatively minor (see Figure 8-8a), and this can be explained in terms of the differences in retention parameters. The maximum normalised release rate to the surface environment occurs shortly after the dose criteria time window, and is about an order of magnitude higher than in the Reference Case, but still almost three orders of magnitude below the regulatory geo-bio flux constraint (Figure 8-8a). The difference compared with the Reference Case is accounted for mainly by the assumption in VS1 of an enlarging defect, rather than by the perturbation to the buffer, as shown in a complementary analysis reported in Section of Assessment of Radionuclide Release Scenarios for the Repository System. Figure 8-7. Evolution of radionuclide release rates from the near field via the F-, DZand TDZ-paths in the Reference Case and in case VS1-BRACKISH.

243 198 Figure 8-8a. Evolution of the geosphere release rates for the VS1 cases and for the Reference Case (BS-RC), with the release rates for each radionuclide normalised with respect to the regulatory nuclide-specific constraints for radioactive releases to the environment. The annual doses to representative persons within the most exposed group (E most_exp ) and among other exposed people (E other ) for biosphere calculation cases BSA-BRACKISH, BSA-HIPH, BSA-HIPH-NF are presented in Figure 8-8b. The dose maximum for E most_exp is msv and the dose maximum for E other is msv for the case BSA-BRACKISH. The dose maximum for E most_exp is msv and the dose maximum for E other is msv for the case BSA-HIPH. The dose maximum for E most_exp is msv and the dose maximum for E other is msv for the case BSA-HIPH-NF. The (typical) absorbed dose rate maxima for plants and animals for the calculation case BSA-BRACKISH is mikrogy/h. This is observed for Reindeer Lichen in terrestrial environments. The (typical) absorbed dose rate maxima for plants and animals for the calculation case BSA-HIPH is mikrogy/h. This is observed for Reindeer Lichen in terrestrial environments. The (typical) absorbed dose rate maxima for plants and animals for the calculation case BSA-HIPH-NF is mikrogy/h. This is observed for Pike in freshwater environments.

244 199 Annual dose [msv] 1.E-04 1.E-06 1.E-08 1.E-10 E most_exp BSA-RC BSA-BRACKISH BSA-HIPH_NF BSA-HIPH 1.E Year Annual dose [msv] 1.E-04 1.E-06 1.E-08 1.E-10 E other BSA-RC BSA-BRACKISH BSA-HIPH_NF BSA-HIPH 1.E Year Figure 8-8b. The annual doses to representative persons within the most exposed group (E most_exp ) (top) and among other exposed people (E other ) (below) for the biosphere Reference Case (BSA-RC), BSA-BRACKISH, BSA-HIPH and BSA-HIPH-NF Cases in Variant Scenario 2 (VS2) In Variant Scenario 2 (VS2) there are no initial penetrating defects in the canisters. Rather, it is assumed that chemical erosion of the buffer takes place due to low ionic strength water penetrating to repository depth, most likely in association with ice-sheet retreat during periods of increased flow rates. Significant buffer erosion is considered unlikely, but cannot currently be excluded in at least some of the deposition holes. Four calculation cases are analysed within this scenario, VS2-H1, VS2-H2, VS2-H3 and VS2-H4, which each consider the failure of a single canister at a given time and location in the repository. Failure within the assessment time frame occurs in those deposition holes that have relatively high near field flows. The four calculation cases analysed for the VS2 scenario correspond to the four canisters that are calculated to fail within this time frame based on a single realisation of the DFN groundwater flow model and a specific

245 200 (reference) set of groundwater flow model assumptions, as described in Performance Assessment. The number and timing of canister failures are regarded as a reasonable illustration, but are not to be interpreted as a precise prediction. Indeed, Appendix 2 of Formulation of Radionuclide Release Scenarios presents arguments that suggest that penetration of low ionic strength water to repository depth will probably not occur, and so there will be no canister failures by corrosion following chemical erosion of the buffer. Currently, however, a few canister failures in this scenario cannot be ruled out. High near-field flows are correlated with low transport resistances in the geosphere facture network. Thus, the deposition holes where canister failures might occur will, on average, be less favourable than most with respect to geosphere transport. However, in some instances, transport paths that leave a deposition hole via a fracture may subsequently re-enter the engineered barrier system and there be subject to much slower flows. This is, by chance, the case for calculation cases VS2-H2, -H3 and -H4 (although not for VS2-H1). As a result, there is no radionuclide release from the geosphere in these calculation cases within the one million year assessment time frame. Figure 8-9 shows the normalised geosphere release rates in case VS2-H1. The main radionuclide contributing to the normalised release is I-129, with smaller contributions from Cl-36 and Se-79, i.e. all long-lived non-sorbing radionuclides. The modelled geosphere release rates show periodic maxima, due to relatively rapid flushing of these non-sorbing radionuclides from the geosphere during periods of high flow (as an ice sheet retreats over the site). The maximum normalised release rate is more than three orders of magnitude below the criterion of one. This analysis considers only one transport path through the geosphere, in one realisation of the geosphere fracture network, and higher peak normalised releases could potentially arise if canister failure occurred in other deposition holes with more rapid transport paths through the geosphere. Transport path uncertainties, as well as the potential effects of irreversible uptake of radionuclides on bentonite colloids and of the potential release of intrinsic colloids in this scenario are considered in complementary analyses reported in Section 12.2 of Assessment of Radionuclide Release Scenarios for the Repository System. Overall, it is concluded that the low peak normalised release rates calculated for a single failed canister in scenario VS2 indicate that the few canister failures that could potentially arise in the more likely lines of evolution (or even the few tens of failures that are calculated to occur in the Performance Assessment if highly pessimistic assumptions are adopted) could easily be tolerated without exceeding the normalised radionuclide release constraint of one.

246 201. Figure 8-9. Evolution of geosphere release rates for VS2-H1, with the release rates for each radionuclide normalised with respect to the regulatory geo-bio flux constraints. VS2-H2, -H3 and -H4 give no release to the surface environment within the one million year assessment time frame. The figure shows calculation over four glacial cycles from 0.6 Ma to 1 Ma (with 3 glacial episodes in each cycle). 8.4 Analysis of the variant scenarios in the surface environment The results for the analysis of all variant scenarios calculation cases in the surface environment are presented in detail in Biosphere Assessment to which the interested reader is encouraged to consult. Here the presentation is limited to the results from analysing variant scenario VS(A), addressing uncertainties in the discharge locations to the surface environment. The annual doses to representative persons within the most exposed group (E most_exp ) and among other exposed people (E other ) for biosphere calculation cases VS(A)-SOUTH1 and VS(A)-SOUTH2 are presented in Figure The dose maxima for E most_exp are msv msv for the cases VS(A)- SOUTH1 and VS(A)-SOUTH2, respectivelly. The dose maxima for E other are msv msv for the cases VS(A)-SOUTH1 and VS(A)-SOUTH2, respectivelly. The (typical) absorbed dose rate maxima for plants and animals for the calculation case VS(A)-SOUTH1 is mikrogy/h. This is observed for Mallard in freshwater environments. The (typical) absorbed dose rate maxima for plants and animals for the calculation case VS(A)-SOUTH2 is mikrogy/h. This is observed for Mallard in freshwater environments.

247 202 1.E-03 BSA-RC VS(A)-SOUTH1 Annual dose [msv] 1.E-05 1.E-07 1.E-09 VS(A)-SOUTH2 1.E Year Annual dose [msv] 1.E-03 1.E-05 1.E-07 1.E-09 BSA-RC VS(A)-SOUTH1 VS(A)-SOUTH2 1.E Year Figure The annual doses to representative persons within the most exposed group (E most_exp ) (top) and among other exposed people (E other ) (below) for the biosphere Reference Case (BSA-RC), VS(A)-SOUTH1 and VS(A)-SOUTH Analysis of the disturbance scenarios in the repository system Three disturbance scenarios are identified that are considered unlikely: accelerated corrosion of the iron insert (AIC); rock shear leading to canister failure (RS); rock shear leading to canister failure followed by buffer erosion (RS-DIL). A set of what-if cases are considered within each of these scenarios assessing the condition of the insert (AIC cases) and time of rock shear failure and buffer erosion Cases in the accelerated iron insert corrosion AIC scenario The accelerated insert corrosion rate (AIC) scenario considers the possibility that an initial penetrating defect in a canister becomes enlarged over time, due e.g. to faster than expected corrosion of the insert, the corrosion products of which will occupy a larger volume than the original metal. This possibility has also been considered in the Monte Carlo simulations and in the VS1 variant scenario. In the VS1 scenario, the hole

248 203 grows gradually (which is more likely than rapid or instantaneous enlargement) between 1000 years and 25,000 years, and the peak release is about an order of magnitude higher than in the Reference Case (Figure 8-8a). The AIC scenario is yet more pessimistic, assuming a much higher than expected insert corrosion rate leading to a sudden (instantaneous) loss of transport resistance of the canister after 15,000 years. The analysis of this scenario focuses on the significance of whether or not a transport path between the canister internal void space and the buffer exists prior to defect enlargement. Two cases have been considered: case AIC-LI (leaky insert) and case AIC-TI (tight insert). In case AIC-LI, the defective canister evolves as in the Reference Case for the first 15,000 years; in case AIC-TI, there is no radionuclide transport path from the canister internal void space to the buffer for the first 15,000 years. In both cases, near-field and geosphere releases increase rapidly from around the time of assumed loss of transport resistance (15,000 years). The releases peak somewhat higher in AIC-LI than in AIC- TI; this can be interpreted as due to the additional release, e.g. of C-14 from corrosion of the other metal parts (fractional corrosion rate per year) to the void space in the canister interior in case AIC-LI before defect enlargement. The evolution of nearfield and geosphere releases after 15,000 years is virtually the same in AIC-LI and in AIC-TI. Figure 8-11 shows the overall geosphere release rates in the two AIC-cases and in the Reference Case, with the release rates for each radionuclide normalised with respect to the regulatory geo-bio flux constraints. In both the AIC-LI and AIC-TI cases, the maximum normalised release rate is about three orders of magnitude higher than in the Reference Case, but remains more than an order of magnitude below the regulatory geobio flux constraint.

249 204 Figure Evolution of the geosphere release rates for the AIC-cases and for the Reference Case (BS-RC), with the release rates for each radionuclide normalised with respect to the regulatory geo-bio flux constraints Cases in the rock shear RS scenario Olkiluoto is located in the Fennoscandian Shield away from active plate margins and is currently seismically quiet. However, the possibility of large earthquakes especially at a time of ice-sheet retreat cannot totally be excluded. The rock shear (RS) scenario considers the possibility of canister failure due to secondary shear movements on fractures intersecting the deposition holes in the event of a large earthquake. Rock shear is assumed to adversely affect the flow-related transport properties of the fracture intersecting the deposition hole where canister failure occurs. The buffer, however, is assumed to continue to fulfil its safety functions. Two cases are analysed: RS1 and RS2, in which rock shear and canister failure are assumed to occur at 40,000 and 155,000 years after disposal, respectively. In both cases, upon canister failure, there is an almost immediate release of IRF radionuclides, such as I-129 and Cl-36, and a more gradual release of e.g. Ni-59, which is released congruently with the corrosion of zirconium alloy and other metal parts. Later, the highest near-field and geosphere release rates are due to Ra-226. Figure 8-12 shows the single-canister normalised geosphere release rates in the RS1 and RS2 cases and in the Reference Case year centred moving averaging has been applied, which is in accord with Finnish regulations (STUK Guide YVL D.5), and this reduces the sharp pulses that occur at times of increased groundwater flow. Peak normalised release rates, with and without averaging, are also shown. The highest normalised peak release rates from the geosphere are in both cases more than two orders of magnitude below one.

250 205 Figure Evolution of the single-canister normalised geosphere release rates in the RS1- and RS2-cases and in the Reference Case (BS-RC) year centred moving averaging has been applied to the RS1 and RS2 curves, consistent with STUK Guide YVL D.5. According to Performance Assessment, up to some tens of canisters could potentially fail in the event of a large earthquake. The overall average annual probability of such an earthquake is around Taking into account that the probability is not constant in time, but is greatest following a period of ice-sheet retreat, and assuming that all canisters that could potentially fail do in fact fail, the peak expectation value of the normalised release rate in the RS scenario (i.e. the peak probability-weighted normalised release rate) is at least around two orders of magnitude below the regulatory guideline. This implies that more than one hundred canisters would have to fail simultaneously before the regulatory geo-bio flux constraint is exceeded. This number exceeds the few tens of canisters estimated to be in critical positions that are vulnerable to failure in the event of a large earthquake (see section 6.3.1). Furthermore, the peak expectation value of the normalised release rate (i.e. the peak probability-weighted normalised release rate) taking into account the uncertain timing of the earthquake leading to canister failure and assuming that all canisters that could potentially fail do in fact fail, is also about two orders of magnitude or more below the regulatory guideline.

251 Case for the rock shear followed by buffer erosion in the RS-DIL scenario In the scenario of rock shear followed by buffer erosion (RS-DIL), the buffer is assumed to undergo either immediate damage due to a rock shear event that causes canister failure or longer-term erosion due to the penetration of low-ionic strength water to repository depth. As the buffer erodes, radionuclides sorbed onto the surfaces of the bentonite are released to the geosphere, either in solution or associated with bentonite colloids. Once sufficient buffer erosion has taken place, advective conditions are established between the canister interior and the geosphere, as in the scenario of corrosion failure following buffer erosion (scenario VS2). RS-DIL is also taken to encompass cases of rock shear where the buffer undergoes more minor perturbation due to the rock shear event (deformation, local thinning). The peak release rates for RS-DIL cases are about an order of magnitude higher than RS cases, but the peak normalised release rate from a single failed canister is still more than an order of magnitude below the regulatory geo-bio flux constraint (Figure 8-13). Taking into account that the probability of rock shear events is not constant in time, and assuming that all canisters that could potentially fail do in fact fail, the peak expectation value of the normalised release rate in the RS-DIL scenario (i.e. the peak probabilityweighted normalised release rate) is around an order of magnitude below the regulatory guideline. The potential effects of irreversible uptake of radionuclides on bentonite colloids in this scenario and of the potential release of intrinsic colloids in this and in the RS scenario are considered in complementary analyses reported in Section 12.4 of Assessment of Radionuclide Release Scenarios for the Repository System. 8.6 Analysis of the disturbance scenarios in the surface environment The results for the analysis of all disturbance scenarios calculation cases in the surface environment are presented in detail in Biosphere Assessment to which the interested reader is encouraged to consult. Here only the results for the human intrusion scenario are presented.

252 207 Figure Evolution of the geosphere release rates in the cases RS1-DIL and RS2-DIL and in the Reference Case (BS-RC), with the release rates for each radionuclide normalised with respect to the regulatory nuclide-specific constraints for radioactive releases to the environment year centred moving averaging has been applied to the RS1-DIL and RS2-DIL curves. The figure illustrates the selection of normalised release rate F A, which is representative of the peak normalised release at the time of canister failure, and F B, which is representative of later peaks at times of ice-sheet retreat. For multiple canister failures, it also shows the probability-weighted normalised release rate, taking into account the uncertain timing of the earthquake leading to canister failure and assuming that all canisters that could potentially fail do in fact fail.

253 Cases in the human intrusion scenario (DS(F)-HI) Scenarios for inadvertent human intrusion caused by borehole drilling have been formulated (see Section 7.3.4). Expectation values of (effective) doses to drilling technicians and site geologists have been derived based on a stylised approach to the dose calculations and estimation of indicative annual probabilities of the intrusion occurring. The peak expectation value of the dose in DS(F)-HI-CANISTER is around an order of magnitude below the regulatory radiation dose constraint for the most exposed people (Figure 8-14). The peak expectation value of the dose in DS(F)-HI-BUFFER and DS(F)-HI-BACKFILL is several orders of magnitude below the regulatory radiation dose constraint for the most exposed people; the detail results are found in Biosphere Assessment. Figure Expectation value of the effective dose for the calculation cases DS-HI-CANISTER-D (top) and DS-HI-CANISTER-G (below), including only the radionuclides that at any time point exceeds a value of 10-6 msv.

254 Complementary analyses Uncertainties in the initial state of the repository system and in its evolution have been taken into account in formulating the base, variant and disturbance scenarios. Complementary analyses have been carried out to further investigate uncertainties related to the model assumptions and parameter values. The complementary analyses include deterministic analyses of complementary calculation cases, as well as scoping calculations, Monte Carlo simulations and a probabilistic sensitivity analysis. The aim of these analyses is to develop a better understanding of the modelled system or subsystems. One key complementary analysis the analysis of more than one defective canister being present in the repository is summarised below. After that, the results of the Monte Carlo simulations and probabilistic sensitivity analysis for a single defective canister are also summarised. These and other complementary analyses are described in full in Assessment of Radionuclide Release Scenarios for the Repository System More than one defective canister in the repository It is possible that there could be more than one canister with an initial penetrating defect in the repository, and more than one of these canisters could be unfavourably located. An illustrative probabilistic model for the reliability of the spent nuclear fuel final disposal canister manufacture and testing has been developed by Holmberg & Kuusela (2011), and this provides the basis for an illustrative probabilistic analysis of the consequences of random emplacement of one or more defective canisters in the repository. To illustrate the consequences, the following quantities have been evaluated for three example key radionuclides (I-129, Cl-36 and C-14): the geo-bio release rate averaged over multiple realisations, where, in each realisation, the number of defective canisters is sampled from the probability distribution given in Holmberg & Kuusela (2011), and these defective canisters are placed randomly in the repository; the probability that the maximum activity release rate from the geosphere as defined above, exceeds the maximum activity release rate from the geosphere due to a single defective canister in the Reference Case. It is shown that the expectation value of the release rate from multiple defective canisters randomly located in the repository is significantly less than the release rate in the Reference Case, in which a single failed canister at a cautiously selected location is postulated. This is because of the low probability of there being many defective canisters and also because the majority of deposition holes have flow-related transport properties that are significantly more favourable to limiting releases than the deposition hole considered in the Reference Case. Furthermore, the probability that the release rate maximum from multiple randomly-placed defective canisters exceeds the Reference Case release rate maximum is low estimated to be about 0.04 %. This provides support for the Reference Case assumption of there being one defective canister in the repository, unfavourably located.

255 Monte Carlo analyses and probabilistic sensitivity analysis Monte Carlo analyses and a probabilistic sensitivity analysis have been performed to assess the impact of uncertainty and variability in model parameters and to identify the model parameters most responsible for the uncertainty in model outputs. Case description and data for Monte Carlo simulations Monte Carlo simulations with 10,000 realisations and probabilistic sensitivity analyses (PSAs) have been carried out for two cases: 1. the hole forever case, where the initial penetrating defect in the canister overpack remains unchanged over time. (The Reference Case can be viewed as one specific realisation of the hole-forever case); 2. the growing hole case, where the defect becomes enlarged over time. This is represented by assuming that the transport resistance provided by the defect is lost entirely and instantaneously after a period that is varied between 5000 and 50,000 years after disposal. The model outputs analysed are the activity release rates from the near field to the geosphere and from the geosphere to the surface environment, normalised with respect to the regulatory nuclide-specific constraints for radioactive releases to the environment given in STUK Guide YVL D.5. To perform the Monte Carlo simulations, probability density functions (PDFs) were assigned to each of the model parameters judged to be affected by uncertainty, several of which (solubilities, distribution coefficients, etc.) are element specific. A total of 160 parameter PDFs were defined in the hole forever case, plus an additional two parameters (Time to loss of hole resistance and Length of canister failed) for the growing hole case. The PDFs were chosen to provide a reasonable representation of the full ranges of uncertainty and variability in the input data. The input data used and the process followed to create the PDFs are presented in Cormenzana (2013b). The parameters sampled are summarised in Table 8-1. Consideration of parameter correlations in Monte Carlo simulations is potentially important as their omission could, in principle, lead to extreme results if unreasonable combinations of parameter values are sampled, which in turn could bias the results of the PSA. This topic has been investigated in Cormenzana (2013b), where Monte Carlo simulations are performed using both correlated and uncorrelated values for all the nearfield flows and geosphere flow-related parameters. The results of the Monte Carlo simulations (expressed as cumulative distribution functions of the peak release rates from the near field and to the biosphere of the main radionuclides and the whole inventory) have, however, been found to be quite similar with and without correlations, and only results without the inclusion of correlations are presented below.

256 211 Table 8-1. Summary of sampled parameters in the Monte Carlo simulations. Details in Cormenzana (2013b). Related to Sampled parameters * Canister failure The spent nuclear fuel Canister interior Buffer Groundwaterbuffer interface Deposition tunnel backfill Near-field flows Geosphere flow parameters Rock matrix Time to creation of transport pathway from canister interior to exterior. Diameter of defect in copper shell. Effective diffusion coefficient in the small hole. Time to loss of hole resistance. Length of canister lateral surface assumed to disappear when transport resistance is lost. Instant release fractions (IRFs) of elements within each of the fuel matrix, zirconium alloy, other metal parts and crud. RN release rate from the fuel matrix, zirconium alloy and other metal parts. Volume of water in the canister interior. Mass of buffer that enters into the canister interior. Solubility of elements inside the canister. Porosity accessible to anions. Effective diffusion coefficients. Distribution coefficients. Solubility correction factors. Solubility correction factors (in damaged rock around the deposition hole) Porosity accessible for anions. Effective diffusion coefficients. Distribution coefficients. Solubility correction factors. Length of tunnel from the deposition hole to the exit fracture. Equivalent water flows through the fracture that intersects: (a) the deposition hole, (b) the deposition tunnel EDZ and (c) the deposition tunnel. Water flux in the tunnel backfill Groundwater travel time, Transport resistance and Length for the F-path, for the DZ-path and for the TDZ-path. Peclet number in the 1-D geosphere paths. Accessible porosity, Effective diffusion coefficients and Maximum penetration depth in the rock matrix. Distribution coefficients. * Note that IRFs, solubilities, solubility correction factors and distribution coefficients (K d ) are each element specific. Results of the Monte Carlo simulations Figure 8-15 shows the total normalised release rate to the surface environment for the hole-forever and growing-hole cases, summed over the F-, DZ-, and TDZ-paths and over all the calculated radionuclides. The results shown are the mean release, the 1st and 99th percentiles, the 5th and 95th percentiles, the median (50th percentile), and maximum values as functions of time. The figure shows that, at any time, the uncertainty in the value of the total normalised release rate is at least of four orders of magnitude. In the hole-forever case, the peak total normalised release rate to the surface environment is more than two orders of magnitude below the regulatory geo-bio flux

257 212 constraint in all realisations. The peak normalised release rate of around in the Reference Case (also shown in Figure 8-15, upper graph) lies between the 95th and 99th percentiles in the corresponding Monte Carlo simulation. This confirms that the Reference Case uses a cautiously selected set of parameter values, giving a peak release rate that is towards the upper end of the range calculated in the Monte Carlo simulation. In the growing-hole case, the peak release rate also remains below the regulatory constraint, although in some realisations the margin is less than an order of magnitude. The model assumption of the growing-hole case that the transport resistance of the defect is lost entirely and instantaneously at a given time is unrealistic and hypothetical. Thus, the growing-hole case, as modelled here, can be considered as a bounding case since the rate of enlargement of the defect over time, if it enlarges at all, is uncertain. (The more realistic assumption of a gradually enlarging defect is considered in the variant scenario VS1, see Section 8.3.1) Figure 8-16 shows the mean normalised release rates to the surface environment in the hole-forever and growing-hole cases, including the contributions of different radionuclides. As in the Reference Case, C-14, Cl-36, I-129 and Cs-135 dominate the release at different intervals within the assessment time window. In both the holeforever and the growing-hole cases, C-14 is the most important radionuclide contributing to the peak release rate, followed by Cl-36 and I-129, with Cs-135 playing a much smaller role. In the growing-hole case, the actinides and progeny are important in the long term, while in the hole-forever case their contribution is negligible. Overall, the Monte Carlo simulations show that radionuclides controlling the total releases from the near field and to the biosphere are those with little or no sorption on the buffer, backfill and rock matrix. Radionuclides with strong sorption on the buffer, backfill and unaltered rock make practically no contribution to the peak total releases from the near field and from the geosphere. These results are consistent with those for the Reference Case.

258 213 1.E+00 Normalised release rate to the biosphere 1.E-01 1.E-02 1.E-03 1.E-04 1.E-05 1.E-06 1.E-07 1.E-08 1.E-09 Total Hole forever Maximum value 95 th percentile 99 th percentile Reference Case Mean Median 50 th percentile 1.E-10 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 Time (years) Regulatory geo-bio flux constraint 5 th percentile Minimum 1 st percentile value 1.E+00 Normalised release rate to the biosphere 1.E-01 1.E-02 1.E-03 1.E-04 1.E-05 1.E-06 1.E-07 1.E-08 1.E-09 Total Growing hole Maximum value 95 th percentile 99 th percentile Mean Median 50 th percentile 5 th percentile 1 st percentile Regulatory geo-bio flux constraint Minimum value 1.E-10 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 Time (years) Figure Total normalised release rate to the surface environment in two Monte Carlo cases with 10,000 realisations (hole-forever and growing-hole cases). The Reference Case release rate is also shown in the upper figure.

259 214 1.E+00 Mean normalised release rate to the biosphere 1.E-01 1.E-02 1.E-03 1.E-04 1.E-05 1.E-06 1.E-07 1.E-08 1.E-09 Hole forever I-129 Rest of fission and activation products Total Regulatory geo-bio flux constraint C-14 Actinides and progeny Cs-135 Cl-36 1.E-10 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 Time (years) 1.E+00 Mean normalised release rate to the biosphere 1.E-01 1.E-02 1.E-03 1.E-04 1.E-05 1.E-06 1.E-07 1.E-08 1.E-09 Growing hole I-129 Rest of fission and activation products Total C-14 Regulatory geo-bio flux constraint Actinides and progeny Cl-36 Cs E-10 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 Time (years) Figure Mean normalised release rates to the surface environment in Monte Carlo cases with 10,000 realisations. Results for the most important radionuclides, the other fission and activation products, the actinides and their progeny, and the total inventory.

260 215 Probabilistic sensitivity analysis (PSA) The analysis of the PSA of the parameters used in the base scenario provides a rich source of understanding of the sensitivity of model output to variation in input parameter values, allowing the most important parameters and parameter combinations to be determined. The most useful of the calculated sensitivity measures have proven to be the rank correlation coefficients (RCCs), which identify the parameters whose uncertainties have the greatest effect on the spread (uncertainty) in the model output. These are identical to standardised rank regression coefficients (SRRCs) when all the uncertain input parameters are independent (uncorrelated), a condition fulfilled for the Monte Carlo simulations described here. Of the graphical methods used, scatter plots and mean rank plots have proved particularly informative. Realisations can be ranked according to the value taken by a given model parameter. The mean rank of realisations in a subset giving the highest 10 % or lowest 10 % values of an output variable (e.g. the peak release rates of a given radionuclide) indicates whether the parameter tends to take high or low values within that subset. Figure 8-17 shows the example of Cs-135 peak release to the biosphere in the hole-forever case as an output variable. The figure shows that the realisations with the highest Cs-135 peak release rate to the biosphere are characterised, for example, by low values of the distribution coefficient of Cs in the buffer and host rock (K d (Cs)) and high values of the defect (small hole) diameter and spent nuclear fuel alteration rate Mean Rank Canister failure Small hole diameter De in the small hole Waste and canister interior Fuel alteration rate Hole forever Q F WL/Q for F-path Q DZ De(Cs) in backfill Kd(Cs) in buffer WL/Q for DZ-path WL/Q for TDZ-path Lowest 10% Highest 10% Peclet number De Kd(Cs) Mass of buffer in the cavity Buffer and backfill Near field flows Geosphere flow parameters Unaltered rock 0 Parameters Figure Mean ranks for all the parameters in the 10 % of realisations with the highest/lowest Cs-135 peak release rate to the surface environment. Using the mean ranks of the (more than one hundred) input parameters that are known to have no effect on the output variable, it is possible to identify the range of values of the mean rank that is not statistically significant: this is the grey band in the figure.

261 Combinations of repository radionuclide release scenarios Possible combinations of radionuclide release scenarios are considered in Section 7.4, and four scenario combinations are identified that require further analysis: 1. The base scenario (BS) in combination with corrosion failure following buffer erosion (VS2) or in combination with the rock shear scenarios (RS and RS-DIL). The analysis demonstrates that the increase in peak normalised release rate that occurs if the canister affected by corrosion failure or failure due to rock shear is assumed to have an initial penetrating defect is more than offset by the reduction that occurs when this peak is multiplied by the very low probability of occurrence of these combined scenarios. The combined scenarios are thus less penalising than the VS2, RS or RS-DIL scenarios alone. 2. Accelerated insert corrosion rate (AIC) in combination with rock shear followed by buffer erosion (RS-DIL). The time of enlargement of the defect, if it occurs at all, is highly uncertain. Analysis shows that to have any impact on the peak release due to the RS-DIL scenario, the enlargement would need to occur shortly (i.e. within a few thousand years) before one of the periods of increased water flow associated with ice-sheet retreat considered in the RS-DIL scenario. The impact of this low probability occurrence would then be limited, dominated by I-129 and Cl-36, since C-14 will have decayed to insignificance. 3. Corrosion failure following buffer erosion (VS2) in combination with rock shear followed by buffer erosion (RS DIL). The peak normalised releases in cases VS2-H1 and RS1-DIL both coincide with the same period of increased groundwater flow associated with a particular ice-sheet retreat period at around 600,000 years (see Figures 8-9 and 8-13). The peak releases can thus simply be summed to give the peak release in the combined scenario. The combined peak release is, however, dominated by the RS1-DIL component, the peak due to VS2- H1 being around an order of magnitude smaller. 4. Accelerated insert corrosion rate (AIC) in combination with corrosion failure following buffer erosion (VS2). The peak normalised release in case AIC-LI is almost two orders of magnitude higher than that of case VS2-H1. The AIC-LI release declines to insignificant levels by the time that the canister fails in case VS2-H1. Thus, combining the two cases does not affect the magnitudes of the peak releases from either. 8.9 Summary of safety assessment results and uncertainties The scenarios analysed have considered a range of events and conditions under which releases might occur. A range of calculation cases has been analysed for each scenario. Case assumptions have been applied within each scenario that include taking pessimistic views on the severity of initiating events, subsequent degradation of engineered barriers and migration paths from defective or damaged canisters. Cautious combinations of scenarios have also been analysed. Parameter uncertainties have been

262 217 investigated most thoroughly for the base scenario for radionuclide release using Monte Carlo simulations and PSA to complement deterministic analyses. Model results are found to be consistent with scientific understanding and indicate robust attenuation and delay of radionuclide releases in the event of canister failures. A set of biosphere calculation cases have been analysed for scenarios taking into account uncertainties in the radionuclide releases discharge locations to the surface environment, the development of the surface environment, the radionuclide transport in the surface environment and in the dose calculations. The results are reported in Biosphere Assessment. In this report only a sub-set of all results are presented Geosphere release rates Figure 8-18 shows the peak normalised activity release rates from the geosphere to the surface environment, and timing of the peak rates, for the base scenario, variant scenario and disturbance scenario cases. All these results refer to failure of a single canister. For cases RS1, RS2, RS1-DIL and RS2-DIL, 1000-year moving averaging has been applied before calculation of the peak rates, consistent with STUK Guide YVL D.5. Figure Peak normalised geosphere release rates for all calculation cases within the base, variant and disturbance scenarios, each assuming the failure of a single canister. Colours are used to group cases by scenario. * indicates that 1000 year averaging is applied, in these cases. The right hand subfigure shows ranges of values for the peak probability-weighted normalised release rates in the RS and RS-DIL scenarios. These ranges arise due to uncertainties in the numbers of canisters failing due to rock shear, as well as in the timing of failure.

263 218 The lowest peak normalised release rates are for the Reference Case (BS-RC) and sensitivity cases within the base scenario. In all cases, peak normalised release rates to the surface environment are below the regulatory geo-bio flux constraint by around an order of magnitude or more. In any of these scenarios, more than one canister may fail. In the case of RS and RS- DIL, quantitative estimates have been made of ranges of values for the peak probabilityweighted normalised release rates (see the right hand subfigure in Figure 8-18). These ranges arise due to uncertainties in the numbers of canisters failing due to rock shear, as well as the timing of failure. They are around an order of magnitude or more below the regulatory geo-bio flux constraint. For the VS2 scenario, arguments are presented in Formulation of Radionuclide Release Scenarios report that suggest that penetration of low ionic strength water to repository depth will probably not occur, and so there will be no canister failures by corrosion following chemical erosion of the buffer. Currently, however, a few canister failures in this scenario cannot be ruled out. The results shown in Figure 8-18 suggest that multiple failures could be tolerated without the regulatory geo-bio flux constraint being exceeded. Other scenarios, including the base scenario, postulate the existence of one or more canisters with initial penetrating defects. The currently available data are insufficient, even when expert judgement is used, to make a reasonable estimate of the probability of emplacing a defective canister in the repository. However, with additional data on the welding process and continued development of the NDT process, it seems practicable in the future to show that the probability of more than one initially defective canister in the repository is less than one per cent. Furthermore, even if there were more than one initially defective canister in the repository, the likelihood that more than one of these would be placed in locations in the repository as unfavourable as that assumed for the defective canister in the Reference Case is very low. Most locations give peak release rates that are orders of magnitude lower than those of the Reference Case. Possible combinations of scenarios have also been considered. Many can be excluded from detailed analysis on qualitative grounds. Where it is appropriate to sum the releases of two different scenarios, the combined geo-bio flux still does not exceed the regulatory geo-bio flux constraint. Monte Carlo simulations, a probabilistic sensitivity analysis (PSA) and a number of deterministic complementary analyses have been performed to obtain a better understanding of the modelled system (see details in Assessment of Radionuclide Release Scenarios for the Repository System). The importance of the properties of any initial penetrating defect in the canister and its evolution over time has been highlighted in these analyses. Quality control and assurance measures have been adopted to ensure transparency and traceability of the calculations performed and hence to promote confidence in the analyses of the calculation cases. These are detailed in Assessment of Radionuclide Release Scenarios for the Repository System.

264 Doses to humans, animals and plants Figure 8-19 shows the annual dose maxima, and timing of the maxima, to a representative person within the most exposed group and a representative person among other people for the biosphere calculation cases presented in this report. The (typical) absorbed dose rate maximum for plants and animals is mikrogy/h in the Reference Case (BSA-RC), observed for Pike in freshwater environment. The dose rate maximum for plants and animals in all the calculation cases presented in this report is mikrogy/h in, observed for Mallard in freshwater environments in the calculation case VS(A)-SOUTH2. 1.E-01 Regulatory dose constraint 1.E-02 Annual dose [msv] 1.E-03 1.E-04 1.E-05 1.E-06 1.E-07 BSA-BRACKISH BSA-RC VS(A)-SOUTH2 BSA-ANNFF VS(A)-SOUTH1 BSA-HIPH BSA-HIPH-NF BSA-TIME 1.E Year Figure The annual dose maxima to a representative person within the most exposed group (E most_exp ) for the biosphere calculation cases presented in this report. 1.E-02 Regulatory dose constraint 1.E-04 Annual dose [msv] 1.E-06 1.E-08 1.E-10 BSA-BRACKISH BSA-RC BSA-HIPH VS(A)-SOUTH2 BSA-ANNFF VS(A)-SOUTH1 BSA-TIME BSA-HIPH-NF 1.E Year Figure The annual dose maxima to a representative person among other people (E other ) for the biosphere calculation cases presented in this report.

265 220

266 221 9 COMPLEMENTARY CONSIDERATIONS AND SUPPORTING EVIDENCE This chapter outlines complementary considerations that provide additional evidence for the long-term safety of disposal according to the KBS-3 concept at the Olkiluoto site. Complementary considerations and additional evidence related to the choice of the geological disposal concept, the robustness of the KBS-3 method and the suitability of the Olkiluoto site are summarised in Sections 9.1, 9.2 and 9.3, respectively. The evidence is fully presented in Complementary Considerations. Results from evaluations of a range of complementary indicators for the repository system are summarised in Section 9.4. The evaluations are fully presented in Assessment of Radionuclide Release Scenarios for the Repository System and Biosphere Assessment. 9.1 Choice of geological disposal The choice of geological disposal as a concept for disposal of radioactive waste is backed by technical experience and international consensus. An appropriately chosen geological formation provides an environment that is stable over many millions of years geological timescales and the nature of changes that can occur is predictable from the geological sciences. A repository concept is developed that is consistent with the chosen geological formation, taking advantage of the beningn or beneficial qualities and designed to withstand expected and unlikely events and processes that could affect the geological formation in the long term. The depth below ground provides buffering of the repository system from processes occurring in the surface environment and protection from unauthorised or inadvertent human actions. 9.2 Support for the robustness of the KBS-3 method The KBS-3 method uses a few simple, common materials copper and iron for the canister, natural swelling clay for the buffer and backfill. This reduces the number of materials whose properties need to be understood and the number of interactions between the materials. Demonstration of safety means that the long-term stability of the engineered barriers must be robustly assured. This has been considered already when choosing materials that are long lasting (natural occurrences/deposits of these materials that have persisted over geological timescales) and with which there is already long experience of their use. The very long timescales of interest much longer than historical experience means that more than empirical experience is required and understanding of the processes involved. For example, the understanding of copper corrosion is based on a combination of experimental evidence and the study of natural analogues. Furthermore, the models used for the interpretation of natural analogues and to make the safety assessment calculations are based on the application of fundamental laws of nature, such as mass and energy balances and the laws of thermodynamics, which provide a robust basis for their use.

267 222 Our experience with natural analogues for both materials and processes provides a high degree of confidence in our understanding of the disposal system and how it will evolve over hundreds of thousands of years (see Chapter 8 in Complementary Considerations). For each engineered barrier and key process, there is increasing analogue evidence to support the models and parameters. Copper is one of the few elements to occur in elemental form as a natural mineral native copper. Although there is no perfect analogue, there is strong evidence from a range of natural occurrences of native copper for very low corrosion rates of copper for millions of years in groundwater and redox conditions similar to, or less favourable than, those at Olkiluoto. Archaeological artefacts, while representing much more variable and often more severe conditions, suggest low localised copper corrosion rates that are not likely to be significant in determining canister longevity compared with generalised corrosion. Taken together, this supports the assertion that the copper canister can provide complete containment for much longer than tens of thousands of years if the surrounding environment maintains the favourable chemical conditions and protects the canister from rock movements. The bentonite buffer needs to maintain its low permeability and plasticity, and prevent microbial activity (which could cause sulphate reduction). Studies of naturallyoccurring bentonite deposits show that mineral alteration processes that are detrimental to the properties of low permeability and plasticity only occur significantly above about C even over geological timescales. Significant changes also depend on a supply of potassium, which will be limited in the buffer (due to the favourable groundwater composition and limited amount of potassium associated with foreign materials to the system). Thus the bentonite buffer will remain stable during the repository thermal period, in which a maximum buffer temperature of around C is foreseen. Several analogue (and many experimental) studies have examined the chemical stability of bentonite under various conditions, but interaction with cement leachate seems to be the only potential detrimental consideration. The use of ordinary cement, either as a component of concrete or cement-based grouts can be avoided by design, substituted by other materials (silica sol grout, for example) or replaced (low ph concrete) so as to ensure the minimum interaction. There are several excellent illustrations of bentonite and other clays functioning as a hydraulic barrier to preserve wood and human cadavers; these analogues also indicate that microbial activity was significantly reduced. One of the strengths of KBS-3 is its simplicity of components and materials. Foreign materials essentially anything that is not part of the EBS introduce uncertainty as they increase the number of chemical components and the number of possible interactions between them. Thus, tracking (and minimising) all foreign materials offers additional support to the robustness of the resulting disposal system.

268 Support for the suitability of geological disposal at the Olkiluoto site In Finland, there are limited choices of deep geological settings, leaving fractured crystalline basement as the only realistic choice of repository host rock. Such rocks, however, have also been considered as suitable for locating a deep geological repository for long-lived radioactive waste in many other countries, including several countries in which alternative rock formations (salt, basalt, tuff and a range of argillaceous sediments) are available e.g. in Canada (see e.g. AECL 1994, p. 3-4 and Witherspoon & Bodvarsson 2006), Switzerland (Nagra 1993) and the USA (Rechard et al. 2011). The important safety functions provided by the host rock and geosphere at a site are: physical isolation due to depth, which minimises the risks of future perturbations (e.g. due to glaciation) and human intrusion; mechanical protection of the engineered barrier system; favourable geochemical conditions (key characteristics being redox, ph and salinity) and sufficiently low groundwater flow rates; and an additional barrier to retard radionuclides, if they are released, due to sorption on fracture minerals and diffusion into the rock matrix. The Olkiluoto site is situated within the Fennoscandian Shield, away from active plate margins. In general, the density and magnitude of earthquakes in Finland is very low; earthquake magnitudes have never exceeded 5 (M=~5) since records began in the 1880s. Further, according to the data from historical earthquakes, the Olkiluoto area is located within a zone of lower seismicity, between two seismically more active belts. There have been only nine recorded earthquakes within 100 km, with the nearest event (M=3.1) at km from Olkiluoto in An important consideration is to find a sufficient volume of rock, with generally low and minor fracturing, to accommodate the spatial extent of the repository. Deposition tunnels must be placed to avoid shear zones or heavily fractured zones, although the access tunnel or shafts may cross such zones. At Olkiluoto, several options have been considered and suitable volumes of rock have been defined such that the deposition tunnels can be placed on a single level at a depth of m. The rock at Olkiluoto is geotechnically suitable for the construction of self-supporting tunnels requiring only light rock support. Water inflows at depth are low and zones of inflow can be treated by local grouting. Significant local experience exists from construction of the ONKALO at Olkiluoto. Evidence from drillholes and the ONKALO show that, in the natural situation, groundwaters at repository depth are reducing, and also have otherwise favourable hydrochemistry (low to medium levels of salinity, chloride and sum of cations). Such conditions are indicative of long periods of negligible groundwater flows and suitable for the longevity and continued functioning of the canister, buffer and backfill. At repository depth, levels of sulphide, which is expected to be the main agent for canister corrosion, and sulphate in the groundwaters that might be reduced to sulphide, are both

269 224 suitably low. There is no evidence that dilute meltwater has reached repository depth during past glacial cycles. The performance of both the EBS and the geosphere as retardation barriers for radionuclides is improved considerably when conditions are chemically reducing. This is because corrosion processes are generally slower under such conditions and many key radionuclides are both less soluble and more highly sorbed. Total salinity can play a significant role in defining the effectiveness of the barrier system. Most information supporting the design and performance of the EBS and radionuclide retention in the geosphere is defined for conditions of low to medium salinity. However, the presence of old, very dense brines provide evidence of geological stability. The ph of groundwater is a less sensitive parameter and is unlikely to have much direct impact on performance of the total system unless it lies at the high (>10) or very low (<4) end of the range. It is therefore advantageous that the host rock has the capacity to buffer any ph excursions caused by either perturbations to the rock itself or directly by the repository. The overall favourability of the Olkiluoto site has been discussed in detail in the Site Description, Chapter 11; the stability of the host rock is also discussed Complementary Considerations, Chapter 7. The characteristics of the site have been determined through the extensive programme of site investigation, and the key features of the site are well understood from site investigation and modelling. Although some uncertainties remain, these are bounded and are allowed for in the safety case. It has been found that the rate of groundwater flow at the planned repository depth is low and geochemical conditions are favourable to the engineered barrier system with reducing conditions, low levels of sulphide and moderate salinity of about g/l. Sufficient volumes of rock will be delineated suitable for the construction of repository panels and deposition tunnels. Evidence that a site like Olkiluoto is appropriate to host a repository can be seen by the presence of numerous ore bodies throughout the crystalline Fennoscandian Shield. Their existence indicates the fundamental barrier properties of fractured crystalline rocks where the geochemical environment of the host rock is appropriate for radionuclide retention as it most certainly is at Olkiluoto. Even at a disturbed site such as Palmottu (see Chapter 7 in Complementary Considerations), penetration of oxidising glacial meltwaters into the site was buffered by the host rock after only a hundred metres depth, indicating the suitability of the crystalline host rock for a repository. Like Palmottu, the Olkiluoto site also shows indications of the presence of glacial meltwaters at depth, but not at repository depth. And also as in the case of Palmottu, meltwaters appear to have mixed with deeper, more mineralised groundwaters or to have been buffered by rock-water interactions so preventing dilute meltwaters reaching the repository horizon. Although data are still sparse (e.g. Pedersen 2008, Pedersen et al. 2010), microbial populations are generally low at the Olkiluoto site, in line with values for most deep

270 225 crystalline groundwaters (e.g. Smith et al. 2001, Fukuda et al. 2010). Additional work in the ONKALO facility will allow this feature to be evaluated more thoroughly in future. 9.4 Safety and complementary indicators Posiva uses the term safety indicator for the quantities derived in safety assessment to assess compliance with the regulatory radiation protection constraints. A range of complementary indicators has been evaluated to highlight the performance of components of the disposal system, and to provide an alternative line of argument for safety. The calculated evolution of the activities in disposal system compartments illustrates where the majority of the activity resides at any given time (Figure 9-1). This demonstrates that the majority of activity is contained within the fuel, zirconium alloy and other metal parts at all calculated times. It illustrates the effectiveness of the waste form and canister in providing long-term containment, even in the presence of a small penetrating defect. Long periods of retention and the slowness of transport processes mean that substantial radioactive decay takes place for the majority of radionuclides before any eventual release to the surface environment can occur. Figure 9-1. Evolution of the total activity, summed over all calculated radionuclides, in the canister, buffer, backfill, geosphere and global biosphere in the Reference Case. Activity in the global biosphere is the time-integrated activity released from the geosphere, taking into account radioactive decay and ingrowth.

271 226 Calculated activity concentrations in the buffer and backfill in the Reference Case have been found to be similar to those that occur naturally in the host rock, although calculated activity concentrations in the VS1 scenario, where enlargement of the defect increases the release rate of radionuclides to the buffer and backfill, are up to around two orders of magnitude higher. Activity concentrations in the buffer and backfill in the Reference Case are also similar to typical values for coal and fly-ash, which are examples of naturally-occurring radioactive material (NORM). Activity fluxes from the near field and geosphere have been compared with the naturally occurring activity flux due to dissolved radionuclides (most significantly, Ra-226) in Olkiluoto groundwater 24. Releases rates from the near field and geosphere in the base scenario have been found to be more than an order of magnitude below the lower bound of the range of uncertainty/variability of Ra-226 flux through the repository area at repository depth. Release rates from the near field are shown in Figure 9-2. Ra-226 and its progeny have the highest dose conversion factor of all radionuclides considered for the stylised deep wells (Biosphere Assessment, Section 6.4) in the biosphere assessment. Thus, the fact that the activity flux from the repository, summed over all radionuclides, is less than the natural Ra-226 flux indicates that the activity flux from the repository is no more radiologically toxic than the natural flux, based on a measure of toxicity (the dose conversion factor s for the hypothetical wells) that is relevant to the site. For the repository variant scenarios, geosphere release rate maxima have been found either to be lower than, or to lie within, the range of naturally occurring Ra-226 fluxes. Only for the unlikely, disturbance scenarios are peak geosphere release rates calculated that are above the naturally occurring Ra-226 flux range. The C-14 released from the repository can be compared with the natural atmospheric C-14 that is taken up by plants and animals in an area similar to the repository footprint. The highest calculated releases of C-14 from the repository to the environment occur when considering, as a complementary case, the possibility of gas-mediated release from the repository. In the case of gas-mediated transport of C-14 from a defective canister through the repository near field and geosphere to the surface environment, the peak annual release rate, averaged over 1000 years, is around a factor of four below the regulatory geo-bio flux constraint (Section of Assessment of Radionuclide Release Scenarios for the Repository System). To put this peak annual release rate in perspective, it is almost an order of magnitude less than the annual uptake of natural atmospheric C-14 by a forest ecosystem over the repository footprint, as demonstrated in Section of Assessment of Radionuclide Release Scenarios for the Repository System. 24 Ra-226 concentrations are only based on four samples taken at Olkiluoto from different groundwater types/depths. These concentrations are relatively high compared with other investigation sites in Finland (e.g. Romuvaara, Kivetty), where at the same depth range Ra-226 concentration is no higher than Bq/L. Nonetheless, the radium concentration is generally higher in the more saline waters which are typical in coastal areas and in Ca-rich groundwaters such as Olkiluoto (Pitkänen et al. 2003, see also e.g. Pöllänen 2003).

272 227 Figure 9-2. Near-field release rates for the various base scenario calculation cases, compared with range of natural Ra-226 flux through the repository area at around repository depth. Finally, it has been found that the performance of each of the repository barriers can often be characterised by two sets of element-specific parameters: transfer coefficients and delay times. In the Reference Case, the lowest transfer coefficients (i.e. the highest resistances to transfer) are from the canister to the buffer, indicating the importance of the assumed size of defect (hole) in limiting releases of the key radionuclides. The assumption of a 1 mm defect, as in the Reference Case and other cases is cautious, and improvements in non-destructive testing methods are likely to reduce the maximum size of defect that could possibly remain undetected.